Research on On-line detection system for natural gas pipeline

HIGH TECHNOLOGY LETTERSI Vol 11 No. 1 I March 200561Research on On-line detection system for natural gas pipelineZuo Jianyong(左建勇), Yan guozhen, Ding guoqing, Fu xiguang, Zhang yunweiSchool of Electronics and Information Technology Shanghai Jiaotong University Shanghai 200030,PRChinaAbstractFour methods for testing the thickness and defect of pipeline are compared and analyzed in this paer. The testing principle of magnetic leakage flux based on electromagnetism is discussed in detail. Fromthe experiments of sensor character the effects caused by some factors are found which give some impotant information for sensor design and this method is proved reasonable and effective. The mechanicaland electrical structures of inspection equipment as well as its working principle and technical featuresare introduced. In this paper, control flow and software design are discussed, too. This detection systemhas been successfully developed. Experiments show that this detection system has high resolution and canKey words: magnetic flux leakage method natural gas pipeline thickness defect, control flowand defect testing of pipeline. They are the MFL0 Introductionmethod2], ultrasonic approach, Electromagnetic AcousticTransducer EMAt )and Remote-Field Eddy CurrentMost countries employ a network of buried pipelines ( RFEC)to transport natural gas from production sites to consumeThe mfl method is a nemethodspectinglocation. These pipelines are always exposed to a wide pipeline. When a ferromagnetic material is subjected to arange of climatic conditions and corrosion stresses and magnetic field irregularities in the structure can causeimage can restrophic failure of the induced flux to leak out into the adjacent space. Intransmission pipelines. So the desire to maintain a reli- the ultrasonic method ultrasonic waves traveling througable supply makes it necessary to inspect them periodical- the pipe wall will be affected by the features they en-for damage caused by corrosion and other faccounter and can be measured to interpret the condition ofOver the years a number of nondestructive inspection the pipe. The most common way of generating an ultratechniques have been developed for detection of damage in sonic wave is to use a piezoelectric device which must bepipeline i. The magnetic flux leakage( MFL )method is in contact with the material to provide a good coupling toan example. Among the existing methods the study on induce an ultrasonic wave. The basic principle of EMATdetection of pipeline wall thickness is seldem seen btis that a conductor carrying current near the surface of ais very important for in-serve pipeline and new pipeline ferrous metal under a magnetic field will induce an elecnspectiontromagnetic force given by Lorentz forces. This combinale inspection equipment for testtion of forces acting on the material will generate Lambing thickness and defect of in-service gas pipeline is built shift and longitudinal ultrasonic waves in the materialin this paper, which is made up of robot and measurement depending on the configuration of the magnet and the diunit. This equipment can freely adjust itself to be fit for rection of the current flow. The RFEC method is based ordifferent pipeline diameters. Besides it can go forward a special physical principle which is the remote-field ednd backward when controlled by industrial computer. dy current effect. Because of this the phase differenceBased on the MFL excited by electromagnetic field the between detecting signal and excitation signal is propormain characters of pipeline inside and outside will be ob- tional to the thickness of pipeline directly. The charactersof four detecting methods are further introduced in Tab. 11 Non-destructive testing method for pipelineH中国煤化工 nat MFL method is better-atecting principle is introCNMHGmodel of thickness meaAt present, there are several methods for thickness surement. The sensor is made up of iron core, coil turnso Supported by the High Technology Research and Development Programme of China( No, 2002AA442110 )and the Province- University CooperationFunds( No 2001 KABAA00A025R密警摁5,2nce should be addressed. E-mailsjtu. edu.cnHIGH TECHNOLOGY LETTERSI Vol, 11 No. 1 I March 2005support arm, HALL element, magnetic concentratingthe flaw. Apart from the reduction of wall thickness ,aquipments and a part of pipeline. A stationary magnetic crack in the pipe wall can also cause the magnetic fltfield is applied inside the pipeline when the sensorleakage. Fig.(b) is a model of defect testing. Whenworking and the flux lines pass through the pipe wall. magnetic flux passes through the pipe wall which has corAlthough most of the flux lines pass through the pipe rosion the direction of magnetic lines will change. At awall a few lines pass through the surrounding media. metal-loss region the flux carried by thin section is lessWhen the pipe wall thickness becomes thin the amount than that carried in the full wall, and flux leaks comeof magnetic flux in air is less than that of the normal from both surfaces of the pipe. In addition the shape ofpipeline wall. The HALL element of sensor could detect the gas-coupled field is also changed. Because the mag-the increase and decrease of flux density. The measure- netic flux leakage is very weak the magnetic concentrat-ment of magnetic flux leakage depends on the material ing devices that can aggregate and uniform magnetic fieldproperties of the pipe wall thickness and dimensions of are usedTab. 1 The characters of four detecting methodsItemsMFIUltrasonicEMATRFECPrincipleThe change of magnetic flux leakageTime differenceTime differenceTime differenceExcitation methodElectromagnet PiezoelectricA四m由electric currentDetecting modeContact withContactStructureComplexComplePipeline materialMoreMeMoreMore difficultDefect recognitionGoodemperature effectTeLittleMeasurement precisionHigh qualitative testing defectHigh, lessTherefore the output voltage of sensor is related to magnetic motive force, Ro is magnetic resistance of ironthe change of thickness and crack. How to discriminate core, Rk is equivalent magnetic resistance of air betweenthem We should observe the detecting information near the magnetic pole, R, is magnetic resistance of pipe partthe testing point. If the signal is saltant it must be do is magnetic flux of iron core, ok is magnetic flux of aircaused by defect. If the signal change is level and between magnetic poles, and g, is magnetic flux of pipesmooth, it must be caused by the thickness change or part. According to the knowledge of magnetic circuit,welarge area corrosioncan draw the conclusions as followsFig1(c) is the magnetic circuit model where F90=k+Iron core Coil turns Support ar MagneticcentrationR=业红中国煤化工CNMHG电PipelFig 1 Schematic of MFL inspection principleHIGH TECHNOLOGY LETTERSI Vol 11 No. 1 I March 200563change only if the material of pipeline changes ; yBR0+R‖RwhichOk= Bk'SkFrom Eg. (3), we can draw the conclusion that Bk(2) will decrease when t increases and this change will beFcome slower and slower. That is to say, high resolutionwill be got when thin pipe wall is detected. The characterof sensor depends on the parameters a B and ywhere Bk is equivalent magnetic flux leakage intensity of 2 Sensor character analysisetween two magnetic poles, Sk is equivalent cross-sectional area of them B, is magnetic intensity ofIn order to study the characters of sensor a testingpipeline part, S, is cross-sectional area of it, n is thenumber of coil tums, i shows exciting current, c and I sensor array is made up of four equi-spaced sensors. Toare constants, t shows the pipeline part thickness and u every sensor, the U shape iron core winded by meandershows the magnetic capacity of pipelinecoils is excited by constant-current source The hall el-Rk>>R,, thus, R, //R=R,i according to the ement used to detect the magnetic flux leakadetecting principle mentioned above ,S,the middle ofore The voltage of hall elementall constants. From Egs. (1)and( 2), the relation be- output will be read by high-accuracy digital universal methe magnetic flux leakage in air and the wall thickter. In this experiment, 12 armor plates have beof pipe part can be described as followscondition The thicknesses of them are103.523.924.585.00,5.44,6.58,hey are all of the same mate-rial( No 45 steel )and the same size( 200 x 300mm.I(3) The effects of lift-off distance magnetic concentrating e-quIpB ktion winding the space between sensors and the time ofI +atmagnetizing are all discussed. Some results are shown ins Ro Ac and b-it ( ,which一鲁- liftoff distarce is2mm-have magnetic concentrating unit--lasungelectrie curent is20A一鲁sc-+liftoff distance is 0no magnetc concentrating unIt4045AThickncss of pipeline wall(mmThickness of pipeline wall (mm)Thickness of pipeline wall/(mm)coil tums mumber is 350→◆ coil turns number sl050Sensor space is Smm-+magnetizing tme is Issensor is 25mmmagnetizing time s 45H中国煤化工CNMHGThickness of pipeline wallmm)Thickmess of pipeline walM(m)Thickness of pipeline walk(mm)Fig 2 Characteristic curve of sensorHIGH TECHNOLOGY LETTERSI Vol, 11 No. 1 I March 2005From Fig. 2( a), we can see that the voltagedifference in magnetizing time only changes the voltage ofHALL output and the sensitivity of sensor are related to HALL output. As we all know that the poor magnetic con-the liftoff distance. Because of the weak magnetic induc- ductivity and strong magnetic resistance will be achievetive capacity of air the magnetic resistance of circuit will because of the outside intensive magnetic field. The longlong liftoff time of magnetizing will cause the increase of the voldistance makes the pipeline part wallof HALL output. During the test, the data acquisitionTherefore short liftoff distance can help to increase thhould be finished at the same time that is to say themeasurement accuracytime of magnetizing is conform. Otherwise, many meaFrom Fig. 2(b), magnetic concentrating equipment surement errors will be obtainedmay not only increase measurement accuracy but also ef-Besides of these we can draw many other concluof high magnetic inductive capacity material, the weak less than 0. ImV/G and the state voltage output h isfect HALL output voltage. Because the equipment is made sions in experiments. I The HALL sensitivity becommagnetic flux leakage can be concentrated and uniformed. comes less than 0.5mV/G when temperature increasesFigs. 2(c) and d ) show the results caused byfrom -25 to 100C. There are few measurement errorsciting magnetic field. Increasing the exciting magnetic because of the different temperature. 2)The defect canfield can improve the sensor sensitivity if the tested pipe be identified easily. Good experience and testing informapart has been magnetized to saturation. The magnetic in- tion is very important to tell the grade of defect. (3)decrease There are more measurement errors because of the differgreatly if the external magnetic field is too strongent material of pipeline. Thus the sensors should be calThe effect of the space between sensors is shownibrated again when different material pipeline is testedFig 2(e). To the same stationary magnetic circuit, larger space will make the pipe part magnetized weakly. 3 Detection systemflux leakand the sensor resolution becomes low. A conclusion we 3. 1 System composition and operating principlecan draw is larger space is of no use for improving sensorFig 3 shows the gas pipeline inspection equipmentresolution, but a thing ofThis equipment is a self-contained unit incorporating apipeline part should be magnetized to saturationnumber of related systems. They are robot,measurementFig2(f) tells us the relation between the time of unit and electrical control systemagnetizing and sensor resolution. We can see that thereentFig 3 Pipeline detection equipmentRobot controlled by industrial computer is a selfdriving unit, which can adjust itself through a parallel i[ esreent mitfour -bar linkameet differentpipeline diameters. This four-bar linkage mechanism is i:xed onbot and controlled by step motor 1( see !Fig 4). A turbine-worm type reducing mechanism con-systentrolled by step motor 2( see Fig 4 )can drive the robot to中国煤化工computermove along the pipeline axisCNMHGThe measurement unit is composed of three partStepmotor IStepmotorThe first is basal disc. The second is 32 wheel finder decoatrol cardStep actorUnderground :! Overgrices that can meet different pipeline diameters by adjust-ing themselves. The last includes sensors array and theirreflex devi t tafre fixed on the basementFig. 4 Electric structure of inspection systemHIGH TECHNOLOGY LETTERSI Vol 11 No. 1 I March 2005Fig. 4 shows the electrical structure of the equip- the sensor calibration curve, collected data and tempera-ment. There are two main parts under ground. One is ture compensation algorithm we can get the thickness ofrobot controlling unit the other is measurement unit con- the whole pipeline. Based on the defects pattern librarytrolling and data acquisition and processing system. They collected data and Neuro-Network method the defects ofboth controlled by embeddeigpipeline can be obtainedthe output of encoder will be read by count card thus theite of robot can be known. Step-motors I and 2 are drivTo initialize the inustrial couterand embedded systemen by step-motor control card in order to adjust themselvesand go forward or backward. In the measurement unitthe procedure of sensor magnetization is controlled by re-lay actuator. Through filter amplifier, the voltage ofHALL output which is about 2-5V will be read by a 3channel 12-bit A/D card and then will be exchanged Itween embedded system and the industrial computer. Besides these, the measurement errors will be compensatedby temperature sensorThis inspection equipment has four main characteristics: 1 The separated control mode between robot andmeasurement unit makes the draft load decrease heavily(2) Because of the contact testing method, high efficiency of magnetization and high sensor sensitivity will be got(3) Because of the wheel finder devices, the measurement unit can go forward and backward, but not rotate(4) Because of the wheel finder devices, the obstacle-Isurmounting performance of equipment is better3. 2 Control flow of detection systemThe control flow of detection system is shown as F5, which includes robot control and measurement unitFig 5 Control flow diagrammodel between robot and measurement unit is adoptedThe measurement unit doesn' t work until the radial ad- 3. 3 Experiment of pipelinejustment tasks have been done and the robot has movedFigs 6 and 7 show the experiment of pipeline. Inor a certain distance. Thus a testing period is over and this experiment, four pipelines are tested by the measure-the next period will begin. In this way the wholepipeline will be inspectedFor software design multilevel interruption programswritten in C language are used in embedded system. According to the need of inspection the software has beendivided into several modules controlled and operated by aninteractive program. The program was written in VisualC++6.0 in Windows in the industrial computer. A lotof things such as task scheduling, program setup, datathrewange and human-machine interaction have been doneWhen a testing period is over the collected data will中国煤化工be exchanged between the embedded system and the in-CNMHGdustrial computer in time. Then the information of thickment unit discussed above. These pipelines are of thesame material( No 45 steel ) the same size( 450mm xware program in the industrial computer. This program in- 2m)and different wall thicknesses(3.75mm 4.50mmthe other i gne is the thickness algorithm program, 5.70mm and 7. 40mm ) From Fig. 7, we can draw thecludes two Dct recognition program. According to66HIGH TECHNOLOGY LETTERSI Vol, 11 No. 1 I March 2005conclusion that the relation between thickness of pipeline(3) The influence of environment temperature to theall and average voltage of HALL output is clear and in measurement errors isn t obviousline with Eq. (3). Besides the measuring unit has high(4)The defect can be qualitatively determresolution and can be made into practicely but it is difficult to give the quantitative parameters(5)The sensor should be calibrated again when different material pipeline is testedThe mechanical and electritructurestecting system as well as its working principle and techflow and software design are discussed too. The inspection equipment has been successfully developed andproved to be put into practice by some experimentsReferThickness of pipeline well/mmS P, Eun S P. Improvement of the sensor system inmagnetic flux leakage-type nondestructive testing. IEEEFig. 7 The relation between the thickness of pipelineTransaction on Magnetics 2002, 382):1277well and average voltage of HALL output[2 I Ivanov P A, Zhang Z, Yeoh C H, et al. Magnetic flux leakage4 ConclusiontransmIssIon pIIEEE Transaction on Magnetics, 1998, 345)3020signed and the experiment work plate is buy or is de- [3 1 Miya K. Recent advancement of electromagnetic nondestructiveBased on the principle of MFl the serinspection technology in Japan. IEEE Transaction on Magnetthe sensor characteristics experiments some conclusionss,2002,382):321[4 ] Zhang Z, Udpa L, Udpa S S, et al. An equivalent linearcan be drawn as hollows(1) It is reasonable and feasible to test the thicknessel for magnetostatic nondestructive evaluation. IEEE Transaction on Magnetics, 1996, 323):718of gas pipeline wall with MFL method(2)In the sensor design magnetic concentrating eZuo jianyong, born in 1977. He received his BSquipments are important to increase the sensor sensitivityand Ms degrees from Southwest Jiaotong University inThe resolution of sensor is not related to the difference of 1999 and 2002 respectively. Now he is a Ph. D. candi-magnetizing time. But many measurement errors will be date in Department of Information Measurement technologygained if magnetizing time is non-conforming. Besides, and Instruments of Shanghai Jiaotong University. His re-shorter lift-off distance, higher exciting current, more coil search interests include robot and nondestructive testingturns and smaller gap of sensors can all help to increasethe sensor sensitivit中国煤化工CNMHG