Home

Abstract

Contents

1. 0 15 or Vi Si C 2 N 7 Wy PIM Pa i co E gt 4 hl dan d ul y 1 y 4 7 Uh H y IN i A uw VALE ju v lt Y d h I Av ul J UNT LW W Ph hv m J gt 0 J J U 0 2 4 6 8 10 12 time min 1 c 9 05 Q gt time min Figure 3 37 Cascade control with W in the inner loop and P2 in the outer loop Miniloop Experimental testing and verification 42 However the valve opening Is still not in the unstable area Further attempts to force the process into the unstable area only resulted in the reappearance of the pressure oscillations associated with slugging Table 3 7 Control parameters Inner loop Outer loop 028bar 1 5 min l T 20 5 Like in the previous case the low gain in the inner loop means that the stabilizing task is left to the outer loop All attempts to increase the gain led to severe oscillations for the actuator and the flow measurement due to the disturbances and noise picture as described in the previous chapter According to storkaas 2 the cascade configuration with mass flow in the inner loop and pressure drop over the choke valve should stabilize the process To prove this a simulation was done using the simplified slug model Attempts to tune the controllers manually were not successful but by using the
2. O aaa Seite 3 Mass Flow Meters Seite 4 Mass Flow Controllers 5 Seite 5 Conversion factors Flow profile sensitivity Seite 6 Principle of Operation Basically the instruments consits of a metal block with a straight bore Two stainless steel probes protrude inside the bore a heater probe and a temperature probe A constant difference in temperatu re AT is created between the two and the energy required to main tain this AT is dependent of the mass flow rate Generally speaking we can say that the higher the flow the more energy is required to maintain the chosen AT which is usually approx 38 C Overall we can state that King s law applies to the relationship between heater energy and mass flow and the follo wing formula can be derived total heater power Po heater power offset at zero flow pz P CO C constant m mass flow n dimensionless number typ 0 5 Today we are working along side more than 20 distributors worldwi de You will find your personal contact on the back page of this bro chure or under www mw instruments com Our intruments are suitable for the use in the pharmaceutical che mical and semiconductor industries as well as in the gas and food industry Of course we are your competent address for special solu tions Pressure drop aaa seite 7 Model number identification seite 8 Technical specifications seite 9 Ultra fast sensor Digit
3. A LO og indstillet v rdi TEACH lt O um Dm rer EA H TI I LU L OFF fabriksindstilling 447 fr TEAGH eane CO fra 0 og op TEACH Skiftende Li PENA man oR L je TEACH K 1 DERES OEH TEACH es gt MODE Vise tal p hovedet funktion Skiftende Standard abrisindstting 19 og ID 3488 OR L 903900 Teach lt 1 Ze Omvendt visning dei cd Note Det er ikke muligt at s tte en nedre gr nse der TEACH lt L er h jere end den vre gr nse E3X DA N OMRON E3X DA N F lsomhedsindstilling Teach SET Mode Der er 4 m der at udf re Teach p Light ON Dark ON omskifter N r Teach er udf rt vil forst rkeren fungere med den indl rte Funktionsm de Indstilling v rdi Hvis displayet blinker er der sket en fejl under Teach og pro ceduren skal gentages Light ON vm Fabriksindstilling Dark ON DEE Indstil mode switch i SET for at p begynde Teach Maximal f lsomhed To punkts Teach med og uden emne 1 S t mod itch SET 1 S t mode switchen p SET Er SET WW SET EMI 1 2 Tryk p TEACH knappen i minimum 3 sekunder 2 Tryk p TEACH knappen i ca 1 sekund n r emnet er i tasteposition TEACH TI lt 3s 3 Teach er udf rt n r niveau displayet skifter fra r d til gr n Niveau displayet vil vise den aktuelle v
4. TEACH t pl E5 2 MODE MODE M M Cancal Execute initial reset Figure 5 4 Resetting the slug sensor to default setting 15 Step 2 Recording the sensor value corresponding to pure water Record the value showing in the sensors digital display when measuring pure water The display will vary between 0 and 4000 The value will depend on the amount of light being returned to the sensor 4000 means all the light has returned as for air When measuring water the display should take a value between 500 and 1500 If the value 15 higher then 1500 the water is absorbing to little light Add more colouring matter as described in chapter 5 4 until the value is within the given bounds Step 3 Setting the lower limit for monitoring The user must then set the lower limit for monitoring a bit higher then the value recorded for pure water This is done to remove the unwanted spikes caused by the surface of the phase transitions If the digital display is showing a value of 1000 for pure water the user should set the lower limit for monitoring at 1200 Seat the mode selector to SET SET Press and hold the mode button MODE until the display shows R LO Then press and hold the teach button until the display show the desired value for the lower limit as shown in figure 5 5 The lower limit will increase in increments of 100 When the desired value is reached release the teach button and switch the mode
5. 22 SAG nn Internal thread Fittings Es What BB G1 4 OE OD 1 2 22 1 4 op _ DD 61 33 6 mm OD E 58 44 12 mm OD FS SIUE ZZ specify 55 1 2 OD i 35 20 uo ID 14 101 min 99 24 20 54 50 ly min Supply Voltage 24 Vdc C 15 Z specify Seals Option 3 Output V Viton 0 5 Vdc EPDM G 4 20 mA P PTFE Elast sourcing 2 specify Style Option 1 Material F controller N C A Aluminium H sensor only 5 55 316 Z specify Z specify Enquiry and Ordering Information In order to supply the correct instrument for your application we request you to state type of gas flow range operating temperature and pressure for controllers supply pressure and back pressure electrical connection desired output signal type of process connec tion and seals Based on this information we perform the following actions calculations 15 100 I min 25 200 I min 45 400 min 55 500 l min 16 1000 I min 26 2000 min 36 3000 min 46 4000 l min 56 5000 min 66 6000 I min 76 7500 I min 99 specify Display Option 5 0 none A flow in B summary Z specify Sensortyp Option 6 S standard FR fast response Bus Connection Option 7 A analog DP digital Profibus DN digital device net DR digital RS 232 DF digital Flow Bus Convert the desired flow to A
6. 12 843 0 0096 0 063 11 0 0096 0 06311 12 913 0 0094 0 06241 0 0094 0 06241 12 945 0 0094 0 062 11 0 0094 0 06211 12 969 0 0093 0 06191 0 0093 0 06191 12 997 0 0093 0 06161 0 0093 0 06161 13 027 0 0092 0 06141 0 0092 0 06141 13 058 0 0092 0 06111 0 0092 0 06111 13 081 0 0091 0 06091 0 0091 0 06091 13 090 0 0091 0 06091 0 0091 0 06091 13 097 0 0091 0 06121 0 0091 0 06121 Q 5 4594 0 1379 0 0453 5 5478 0 1456 0 0424 5 5836 0 1484 0 0415 5 6002 0 1498 0 041 I 5 6131 0 1507 0 0407 5 6274 0 1518 0 0404 5 6430 0 1530 0 0401 5 6593 0 1542 0 0397 5 6713 0 1551 0 0394 5 6756 0 1556 0 0394 5 6776 0 1574 0 0392 Table A 2 0 1 0 15 0 18 0 20 0 22 0 25 0 3 0 4 0 6 0 8 Appendix A Valve opening z Poles for the system at different valve openings poles 9 2653 0 0246 0 14591 0 0246 0 14591 9 5311 0 0133 0 18621 0 0133 0 18621 9 7067 0 0030 0 21571 0 0030 0 21571 9 8250 0 0047 0 23511 0 0047 0 23511 9 9468 0 0131 0 25431 0 0131 0 25431 10 1399 0 0268 0 28231 0 0268 0 28231 10 4951 0 0522 0 32651 0 0522 0 32651 11 3499 0 1100 0 40421 0 1100 0 40421 13 6612 0 2365 0 51591 0 2365 0 51591 16 7605 0 3561 0 58071 0 3561 0 58071 20 6258 0 4565 0 61961 0 4565 0 61961 55 Append
7. 0 01 mm 3 mm dia e Lang E3X DAL N E32 T21L 25 mm taste 01 eg afstand 180 ia E3X DAL N E32 T22L 180 E3X DA N Appli Egenskaber kation Lang taste M14 med linser ideel til eksplosionssikre OMRON V rdier i parantes ved anvendelse af E39 F1 linse E3X DA N Standard objekt 2 mindste uigennem sigtige emne Tilladelig b jnings radius Tasteafstand mm 10 mm dia 0 01 mm SEE 20 000 1 20 000 1 afstand applikationer dia 9 800 M E3X DAL ON M3 mulighed for vinkelaftastning med E39 F5 vinkellinser E3X DAI E3X DAB11 N E3X DAL N M fo M M 2 mm dia sm emner Tors 90 mm 40 1 2 mm dia E3X DAI E3X DAB11 N E3X DAL N 01 2 Bojeligt metal fiberror i enden E32 TC200B4 90 mm 40 mm o 9 mm dia qu Lap M3 E32 TC200F4 0 9 mm Bojeligt metal fiberr r i enden Fleksi bel fiber E3X DAL Kan b jes som en ledning B jningsradius 1mm 14 ames dite E3X DALI 4 E3X DAB11 N 75 550 45 350 3 3 E32 TC200 280 2 100 100 700 E32 TC200A 25 mm A E32 TC200E 12 2 22 E32 TC200B E32 TC200B4 E E3
8. 0 4 P d Real part of the worst pole 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 1 valve opening Figure 3 32 Real part of the worst pole Experimental testing and verification 35 3 8 Anti slug control In this chapter different control structures will be tested on the Miniloop and they will be compared to simulations performed on the simplified slug model The criteria for satisfactory control is to stabilize the system at a valve opening that would normally result in severe slugging in open loop 3 8 1 Control with upstream measurements According to the controllability analysis the best choice of measurement would be the upstream pressure P1 A simple PI controller with gain K 22 bar and integral time T 10s will stabilize the system as shown in figure 3 33 The system starts from a state of severe slugging and the controller is turned on after two minutes After an additional two minutes the controller is set to manual and the slugging reappears The top chart in figure 3 33 shows the upstream pressure P1 vs time The blue line 1s the pressure and the red line 1s the set point for the pressure controller witch was set to 0 115 barg The controller stabilizes the system quickly if a bit aggressive The stabilized system experiences small pressure oscillations however these are small compared to the amplitude of the slugging so the tracking performance of the controller is considered as good The actuato
9. 400 500 800 1 000 1 500 2 000 2 500 3 000 3 500 4 000 4 500 5 000 6 000 7 500 ltr ltr ltr ltr Itr ltr ltr ltr Itr Itr Itr Itr ltr ltr For further information please refer to the second table set out below n A E EE uuu 0 5 eee NM EEE 1 6210 mm S8 HN NNNM 26230 Ss Ss E Ss Ss D6250 m EN RE V Pm RE 19 2 BGO EEHEHE PEHE EE E E m E c IEEE M NM NM NU gt 1 gm 100 es OO Cm AP mbar 1 0 10 15 20 25 40 50 60 ltr ltr ltr Itr Itr Itr ltr Itr flow air l min 70 80 90 100 200 300 400 500 800 1 000 1 500 2 000 2 500 3 000 3 500 4 000 4 500 5 000 6 000 7 500 ltr ltr ltr Itr Itr ltr ltr ltr ltr ltr ltr ltr ltr ltr Itr Itr Itr Itr Itr ltr M odel Number Identification Options and Model Numbers M ass STREAM 8 D NNNN Base model D 511 low flow D 512 medium flow D 519 specify D 621 2 mm D 623 8 mm D 625 16 mm D 627 32 mm D 628 56 mm D 629 84 mm Function 0 Mass flow meter 1 Mass flow controller AAA AA Gas connection in out Flow rate Air Option 4 Option 2 12 0 1 l min
10. Appendix A ST A 4 Parameter estimation The liquid flow estimate l min 1s based on the two following equations k z JAP A 4 Q f z JAP p A 5 Method a Linear parameter estimation This method uses equation A 1 There are two parameters k and n and three variables Q z and AP The idea is to rearrange equation A 1 so that the two parameters can be estimated Since only water is present in the system p 1 reducing A 4 to Q k z VAP A 6 By taking the natural logarithm on both sides the equation can be rearranged to In Q In k nIn z In AP A 7 Further rearrangement gives In Q In k n In z A 8 This 1s a normal first order equation with two variables and two parameters XM Py GES A 9 Where x IN Q gt 1n AP x ln z p In k p n By plotting x vs the two unknown parameters can be estimated p will be the intersection with the y axis and p will be the slope Appendix A 58 e Datapoints Linear Datapoints er lt H Figure A 14 Linear parameter estimation plot The first order equation fitted to the data points 1s y 0 5665x 4 2264 This gives k e 68 4 n 0 56 Resulting in the following valve equation 68 4 7279 JAP A 10 Method b Fitting the function f z This method is based on equation A 5 The density term is removed since its for water
11. f z NAP A 11 A simple rearrangement gives Q JAP f z A 12 Appendix A 59 By plotting Q 4 AP vs Z in figure A 14 and fitting a 5th order polynomial forced through origo to the data points gives the following estimate of f z e Datapoints Poly Datapoints 0 1 0 15 0 2 0 25 0 3 Valve position 2 Figure A 15 Estimation of f z The fifth order polynomial fitted to the data using the least squares method 15 f z 70223z 613502 19191z 2705 4z 229 44 A 13 Exponential and logarithmic fits were also attempted but A 13 gave the best results Equation A 13 inserted in A 5 gives the valve equation used to estimate the liquid flow Appendix B USER MANUAL Appendix B 60 USER MANUAL For the lab scale MINILOOP Table of content Elles ches seats cence ee 3 2 Mmuioop and sp 4 o Operating the M OD EE 7 3 1 Start up and shut down procedures reessione E AR 7 5 2 VE 8 3 3 ACUVE CONTO sene 9 SI MONO serer 9 4 The Miniloop block 10 4 1 KE NNN 10 22 Milers and di pare 11 4 3 CASE structure and the odori esiti ca eso tragos co bao rotes dad 12 Was the Cala toy vvs 13 S Ma amenante erenn 14 5 1 FM 14 5 2 DULCE TEE 14 9 3 ETT 15 5 3 1 Calibrating the slug Ee EE 15 5 3 2 Troubleshooting the slug sengor 17 5 4 CM Eeer 17 6 R
12. I S 23 110 Heater oe Flow If even faster times are required then we would recommend using this sensor in conjunction with our digital pc board Further infor mation is available within the section headed Digital version Dimensions as standard s p 4 5 Model number identification s p 8 MUN Wd amp Readout Systems with integrated Power Supply mm G eneral This series comprises standard types for use with analog mass flow meters and controllers The most commonly used functions are offered in compact single zm Functions Power Supply for MFM M FC Indication of flow rate mum Flectrical data Power supply 110 or 230 Vac 50 60 Hz Suitable for connection of instru ments with output signal 0 5 Vac zs M number identification channel table top housing DIN panel mount cassette and multi channel ver sions in 1 2 19 or 19 table top or rack housing Totalization Setpoint potentiometer Sub D socket for instrument connection Max power 24 Vac 0 5 per channel Loge Housing Do 1 2 19 tabletop 42 TE s Uc semen 19 tabletop DM HE ces Co E DM Bio M c Table top cassette TE ape ge te anne Dale Panel mount cassette La e n Code A Suppyvoltage mm nu 0 100 240 mut i DA 20 110 Vac 43 Rearpanel blind D 15
13. The real part of the worst pole has been evaluated and plotted against the valve opening in figure The poles start in the LHP and move over to the RHP when the valve opening is z 0 19 This corresponds with the bifurcation diagram where the system is unstable for z gt 0 19 RHP poler gives a lower limit on the bandwidth for the process This means that the lower limit for the bandwidth will increase as the valve opening increases The system can be stabilized by using feedback control to move the poles RHP zeros results in inverse response and imposes an upper limit on the bandwidth for the process To obtain stability with a satisfactory performance the following is required Wc gt 1 15 zn gt 2 3 p 3 6 Table 3 3 System poles Valve opening L I RHP pole length z 0 15 9 5311 0 0133 0 18621 0 0133 0 18621 0 1867 z 0 3 10 4951 0 0522 0 32651 0 0522 0 32651 0 3306 The different measurements available are listed below Table 3 4 Available measurements Measurement Unit Description Upstream pressure feed inlet Downstream pressure Pm Kg m Density W Kg min Total mass flow Q l min Total volume flow The only upstream downside measurement is the pressure P1 All the other measurements are upstream topside measurements The zeros for the different measurements are given in table 3 3 Table 3 5 Zeros for the different measurements at the operating point z 0 3 Pm W
14. control of the system using only downstream measurements were grim The noise picture and nature of the slug flow prevented the inner loop from stabilizing the system when the mass flow was used The reason it was possible to stabilize the system lay m the stabilizing property of the outer loop In this case the pressure drop over the choke valve P2 is unsuitable for a stabilizing controller according to the controllability analysis The same noise picture that created the problems for the previous case would still be present for this control configuration However an attempt was made using the same tuning strategy as for the case in chapter 3 8 4 The system was brought to steady state in the stable region and gradually forced towards the unstable area by changing the set point and tuning the parameters Figure 3 37 shows the response obtained with the tuning parameters in table 3 7 The process starts in open loop with sever slugging and the controller is turned on after 70 s with a set point of 0 02 barg The controller first stabilizes the system by bringing it into the stable area The straight line in the bottom chart is the valve opening z 0 19 corresponding to the bifurcation point in figure 3 29 Valve openings below this value are in the stable area and visa versa As can be seen from the chart the pressure is slowly brought to its reference value as the valve approach the valve opening z 0 19 corresponding to the set point
15. r reset Digital 12 bit aktuel v rdi gt Digital 12 bit indstillet v rd Digital procent aktuel verd Digital procent aktuel v rd Aktuel analog v rdi Aktuel analog v rdi E3X DA N OMRON E3X DA N Nem at anvende Indstillinger i SET Mode Mode switch SET 74 7I 07 Maksimum felsom dd 0 OC OC OC OC TEACH lt L hed fabriksindstil SET ME Digital visning fra 0 til 4000 ling Der kan v lges Skalering af udgang Tilg ngeligt pa ingsm der Se E3X DA21 N efterfalgende E3X DA51 N sider e E3X DA7 E3X DA9 B 1 gt gt p lastem de Vist som F MODE 25 gt OD vre gr nse Skifter mellem A UP og indstillet v rdi Standard fabriksindstilling at TEACH lt 1 fabriksindstilling iftende Super lang afstand JF d ob t P e J 4 H TEACH I enheder af 100 fra e Wim 4000 og ned Super high speed NN Skiftende react CJ oR UD 9 H 5 uw TEACH 0 fabriksindstilling o Li ob UH lt ali 2500 Justerbar i af 1 ms TEACH lt l fra 0 til 20 ms off i enheder TEACH af 5 ms fra 20 ris og op KL mr 200 3 Ar C LU Skiftende Ge DH ID wes iin Due Eh s 1881 Note Det er ikke muligt at s tte en ovre gr nse der er mindre end den nedre gr nse Teach lt 1 MODE Nedre gr nse Skifter mellem ON
16. 0 20 40 60 80 100 120 time 5 Figure A 12 Open loop data for z 0 14 7 Valve opening 0 16 0 14 S 0 12 a 0 1 0 08 0 06 0 20 40 60 80 100 120 time s 0 15 D g 0 1 N oO 0 05 0 0 20 40 60 80 100 120 time s Figure A 13 Open loop data for z 0 07 Appendix A A 3 Generated data from the simplified slug model Table A 1 shows the different zeros for the measurement alternatives at different valve openings The corresponding system poles at the same valve openings are located in table A 2 Table A 1 Z Pl 0 1 0 1520 0 15 0 1597 0 20 0 22 0 25 0 3 0 4 0 6 0 8 0 1626 0 1640 0 1650 0 1661 0 1673 0 1686 0 1695 0 1700 0 1718 2 0 5239 0 89201 0 5239 0 89201 0 7002 0 79151 0 7002 0 79151 0 7668 0 73941 0 7668 0 73941 0 7969 0 71251 0 7969 0 71251 0 8197 0 69071 0 8197 0 69071 0 8449 0 66481 0 8449 0 66481 0 8720 0 63471 0 8720 0 63471 0 8996 0 60121 0 8996 0 60121 0 9198 0 57451 0 9198 0 57451 0 9269 0 56581 0 9269 0 56581 0 9301 0 57061 0 9301 0 57061 Pm 0 0532 0 01071 0 0532 0 01071 0 0812 0 0312 0 0874 0 0273 0 0899 0 0259 0 0918 0 0248 0 0937 0 0237 0 0958 0 0226 0 0979 0 0216 0 0993 0 0209 0 0996 0 0207 0 0979 0 0209 Zeros for the different measurements at different valve openings W 12 670 0 0100 0 06501 0 0100 0 06501
17. CN11 udgang E3X DA6 E3X DA8 dd ER E3X CN12 Treet ny E3X CN21 ON OFF udgang E3X DA7 E3X DA9 udgang me Analog udgang ie E3X CN22 Udseende Udgang NPN gt MN PNP udgang Standard model L ON OFF udgang E3X DA14V E3X DA44V OMRON E3X DA N E3X DA N Stik med kabel til BUS model Bestilles separat Enhed Udseende Kabel l nge Antal ledere _ ee Komplet tilbeh rspakke for tr dl s programmering er inkluderet Bestykning af fiberforst rker med BUS funktion Eksempel hvis man nsker BUS forbindelse mellem 5 forsteerkere Stik med kabel Master forbindelse Slave forbindelse Standard model E3X CN11 3 leder E3X CN12 1 leder Analog udgang model E3X DA7 E3X DA9 E3X CN21 4 leder E3X CN22 2 leder Bestykning ved 5 s t Forst rker enheder 5 stk Master forbindelse 1 stk Slave forbindelse 4 stk Fiber enheder Separat sender modtager 2 Indikerer modeller hvor det er tilladt at forkorte fiberen ved hj lp af den medf lgende saks E Super lang tasteafstand I Standard mode Super high speed mode Appli Egenskaber Udseende Forst rker Tasteafstand mm Standard Tilladelig kation V rdier i parantes ved objekt 3 b jnings anvendelse af E39 F1 linse mindste radius uigennem sigtige emne OG Lang E32 r 25 mm taste ING afstand 7 ia E3X DAB11 N E3X DAL N 1 660 1 4 mmdia E32 1 330
18. D 11 aer JE H T E rmm uml m ad mj gg gj it ag E Ts SS re A E LS i D 14 Tel ee eee e eee ME De 0 unea A al RG PES Distributor Flow Teknikk as Olav Brunborgsv 27 Postboks 244 1377 BILLINGSTAD Tlf 66 77 54 00 Fax 66 77 5401 E post mail flow no www How no ME m menis OMRON Digital fiberforst rker E3X DA N Brugervenlig og med tydeligt display B V lg imellem 3 tydelige display visninger Digital 12 bit visning 0 til 4000 Digital procentvisning Analog visning B Multifunktion og stort modelprogram Bl a bl LED BUS typer for nem fortr dning samt M8 IP66 model B Beskyttelse mod gensidig interferrens ZH B Skal rbar 1 til 5v analog udgang _ smi Programmeringsenhed B Giver mulighed for fjernbetjent programmering Copy og paste funktion og 10 hukommelsesbanker m Afl sning af parametre beskytter mod uhen sigtsm ssig betjening Typeoversigt Forst rkerenheder Med kabel Udseende Udgang Moder NPN udgang PNP udgang Standard model model S ON OFF udgang E3X DA11 N E3X DA41 N Med aa udgang ON OFF udgang E3X DA21 N E3X DA51 N adii Analog udgang M rkeaftaster med M ON OFF udgang E3X DAB11 N bl LED a Til BUS forbindelse Bestilles separat NPN udgang PNP udgang Standard Sa E3X
19. ON H ndelse OFF Udgangsindikator ON orange Udgangs transistor Belastning rel ON OFF OFF mellem brun og sort Soe Dark ON OFF ON OFF mellem brun og sort OMRON Light ON Dark ON Light ON Dark ON Light ON Dark ON Light ON Display Udgangsdiode orange E3X DA N Sort main circuit Control output Belastningsmodstand 10 kQ min Analog udgang 1 til 5 V OMRON E3X DA N E3X DA N Drift Display visninger i RUN Mode Zero reset RUN Mode Mode switch i RUN Digital 12 bits visning fra 0 til 4000 Mode switch i RUN Digital 12 bits visning fra 0 til 4000 Ce RUN Fabriksindstillet p RUN 25 gt gt move Digital procent visning fra 0 til 999 Reset til 0 visning igen TEACH Vend tilbage til normal visning TEACH MODE K Hold begge taster nede i 3 s Note Zero reset kan udf res s ofte man nsker Analog visning Reset til fabriksindstilling SET Mode Mode switch i SET Manual indstilling i ADJ Mode Piletaster anvendes til fin indstilling Manuel indstilling eller fin indstilling efter Teaching TEACH MODE NA Hold begge taster nede i 5 s TEACH ADJ TEACH CI MODE A C M m F lsomhed skrues op F lsomhed skrues ned ont d oMES MODE MODE Visning i ADJ mode er forskellig fra visning i RUN mode Y RUN mode ADJ mode Afbryd reset Udf
20. SOMES D 6200 Inrush current 250 MX mm Ges No flow 75 aaa 5 5 _ 100 flow 175 222 Control valve if applicable 250 mA max Outputsignal d 0 5 Vdcor4 20 Ultra Fast Sensor Fast response version M W Instruments Mass Flow Meters and Controllers are used for a wide variety of applications and for this reason the sensor has been set up with smooth response characteristics thereby avoiding any possible overshoot of the setpoint There are however applications where the response time of the sen sor or the control valve is the decisive factor and for these applica tions M W Instruments has developed a sensor with the following features Response time 51 up to lt 100ms When using this sensor in connection with a flow controller the fol lowing response times are possible Response time 51 up to 1 5s Digital version All our standard products are equipped with an analog pc board a feature that ensures that they are very economically priced However our well thought out modular system allows us to offer a digital pc board as well thereby giving the options of analog volta ge or current output together with Profibus DP Device Net RS 232 or Flowbus protocols Flow Controller response times down to 0 5 sec and less are possi ble when using the fast response sensor eo Seals Body o Heating System aS 3 d p
21. Tor 2 E vvs 49 Opendoop data Tor z 0 NG 49 Open loop data Tot 023 edente ten s vc eis Cebu bes Vie uod ven 50 Oper loop Gala E EE 50 Open loop dala TOr z EE 51 Open loop Tor 2 0 nesa 51 Open loop data for 0 19 ae 52 Open loop Gata Tor Z OS save Sene 52 Open Toop data TOP Z Vasset 53 Open loop NTT A Ve 53 Linear parameter estimation A e E 58 FIG I vit 61 Introduction 4 1 Introduction 1 1 Background The diploma thesis brings the education as a chemical engineer at the Norwegian University of Science and Technology NTNU to a close It has been carried out at the Department of Chemical Engineering The title for the thesis 1s Anti slug control Experimental testing and verification and can be considered as a continuation of the project Anti slug control for a two phase flow Experimental verification by B rdsen 3 1 2 History Multiphase pipelines connecting remote wellhead platforms and sub sea wells are a common feature of offshore oil production in the North Sea and the signs are that even more of them will be laid in the coming decades 8 This makes the problem connected to multiphase transport of gas oil and water an increasingly important topic for the offshore oil industry Underwater installations allow the untreated well streams from different well cluster and wellhead platforms to be collected and transported into the production pla
22. a higher set point the corresponding valve opening would be in the stable area of the bifurcation diagram figure 3 29 At this valve opening the system would already be stable in open loop When the controller was turned on the actuator would operate in the stable area and the process would stabilize regardless of the parameters used The nest step was to tune the controller so that the pressure was brought to its reference value still in the stable region Then by slowly lowering the set point the process was forced towards the unstable area When the set point was low enough the valve opening would start to operate in the unstable area above z 0 19 In the start the process would start to oscillate at this point but by tuning the parameters the process was slowly brought into the unstable area The biggest challenge lay in the inner loop gain It was evident that the gain had to be increased further if the inner loop were to take over the stabilizing task All attempts to increase the gain further resulted in an oscillating behaviour where the actuator and the flow estimate would oscillate between min and max values The problem lay in the noise and disturbance in the flow measurement Different filters were tested as the tuning progressed but combined with the flow pattern described in chapter 3 4 1 the noise picture made it impossible to stabilize the system with the inner loop The current parameters used in table 3 6 allowed the process to opera
23. axis When both RHP poles and zeros are present in the same system the given upper and lower bounds on the bandwidth will make the stabilizing of the system impossible In order to fulfil the limitations imposed on the bandwidth and achieve a satisfactory performance and resilience the following is demanded Zn gt 2 4 p 2 7 Experimental testing and verification 10 3 Experimental testing and verification 3 1 Apparatus Figure 3 1 shows an overview of the lab scale Miniloop that was used during the experimental face of this thesis The Miniloop was originally constructed by Bardsen 3 as a part of his fifth grade project with the Department of Chemical Engineering at NTNU Some changes have been made to the original Miniloop including purchase and installation of additional equipment These changes will be addressed later on in this chapter FF modules Airinn 5 BT be e PU Figure 3 1 Flow sheet for the Miniloop s can be seen from the figure the Miniloop has a water WT and an air source The water is pumped from the reservoir into the system while the air is let into the system from a pressurized air outlet in the wall The flow rate of water and air is controlled by manually adjusting valves V1 and V2 The pipeline system is constructed of several connecting sections of transparent plastic tubes The pipeline is meant to imitate the pipeline topography where gravity induced slugging occurs wh
24. ca 50G 3 gange i hver retning X Y og Z Materiale Hs PBT Polycarbonate Engelsk Japansk instruktionsvejledning E3X DA N OMRON E3X DA N Grafiske Data E3X DA Parallelt aftastningsomr de typisk ved max f lsomhed E32 TC200 E32 T11R E32 T11 Sender modtager Sender modtager Sender modtager Super long GJE high Afstand m speed T Super lang afstand Super lang afstand Driftsomr de typisk standardobjekt aftastet ved max f lsomhed E32 DC200 E32 D33 E32 L25L Diffus reflektion Diffus reflektion Begr nset diffus reflektion Y Super lang afstand high speed Afstand Y mm Afstand Y mm E3X DAB11 N Parallelt aftastningsomr de typisk Ved max f lsomhed E32 TC200 Sender modtager E32 TC200 E39 F1 Sender modtager 80 60 40 mu FN J REN 20 standard E Super high speed m Super high Standard Super lang f speed afstand Driftsomr de typisk Standardobjekt aftastet ved max f lsomhed Afstand Y mm E32 DC200 Diffus reflektion E32 CC200 Diffus refleksion E32 D11L Diffus reflektion Afstand Y mm OM SN KNE 100 10 SN Pen mm tandard Super lang afstand Super high 30 speed E3X DA N Drift Udgangskredslob U
25. fixed in the model it was impossible to fit both amplitude and frequency P1 barg P2 barg P1 barg P2 barg Experimental testing and verification 32 0 16 0 12 0 08 0 14 0 06 0 0 2 0 4 0 6 0 8 1 1 2 1 4 1 6 1 8 time min 0 04 0 03 0 02 0 01 time min Figure 3 30 Open loop behavior for the miniloop z 0 3 Y Au e d sg 0 0 2 0 4 0 6 0 8 1 1 2 1 4 1 6 1 8 0 16 0 14 0 12 0 08 L 0 11 FN K j N N 2 0 06 0 0 04 0 03 0 02 0 01 Figure 3 31 0 2 0 4 0 6 0 8 1 1 2 1 4 1 6 1 8 time min V DE 0 6 0 8 1 1 2 1 4 1 6 1 8 time min 02 0 4 Open loop behaviour for the simplified model z 0 3 Experimental testing and verification 33 3 7 Controllability analysis This analysis is based on a linearized model around a typical unstable operating point The operating point chosen is z 0 3 The system is unstable for this valve opening in open loop since there 1s a complex pair of poles in the RHP table 3 3 To compare it with an operating point that should be in the stable area according to the bifurcation diagram the poles for Z 0 15 are also included in the table As can be seen these poles are in the LHP hence making the system stable The poles and zeros for other operating points can be found in appendix A 3
26. gets high enough it will push the slug up through the raiser This can be seen when the value drops from 5 to 1 volt The air will eventually penetrate the liquid witch 1s represented by the oscillations between 1 and 5 volts The liquid then falls back to the low point causing the value to increase to 5 again before the cycle repeats itself Experimental testing and verification 24 The nature of the slug flow creates some restrictions and limitations witch reduces the effectiveness and accuracy of the slug sensor In normal gravity induced slugging the gas would penetrate the liquid as bubbles in the liquid flow This would allow the optical sensor to estimate the fraction of water passing the sensor giving us the hold up However conditions in the raiser produced a flow regime during closed loop where large volume of gas flowed as a separate phase in between the liquid bulks figure 3 21 Slug sensor Figure 3 21 Slug flow pattern in the pipe The slug sensor will therefore only be able to indicate the presence of water or air not a m1x of both The oscillations figure 3 20 are therefore a result of this flow pattern and the disturbances caused by the phase transitions By adding a frequency filter in LabVIEW the measurement was improved further figure 3 22 by removing some of the disturbance and averaging the data Figure 3 22 Final slug sensor readings slug se 1 Il TITT Dn LJ a To transform
27. min l Ti 30 The higher value for the gain in the inner loop proves that the stabilizing work is performed by the inner loop The tuning parameters in table 3 8 were also tested on the Miniloop The controller acted by closing the valve resulting in an increase in pressure The pressure eventually became so high that the experiment had to be aborted for safety reasons The simplified slug model continues to show its similarity to the experimental data Even though the cascade controller failed to stabilize the slugging in the unstable area the response 15 similar In both cases the controller brought the system into the stable area before it attempted to bring the pressure to its reference value by bringing the system into the unstable area 3 9 User manual As a Stage in documenting the work done on the Miniloop during this thesis a user manual has been written The manual is considered as a part of this thesis and it can be found in appendix However it is written as a separate report so it can be used independently Because of this the user manual contains some of the information reported in chapter 3 Future work 44 4 Future work e The noise picture and disturbances associated with the flow measurement made it impossible to stabilize the process with mass flow in the stabilizing loop If a more accurate measurement of the flow could be estimated the cascade configuration in chapter 3 8 5 might have worked o The flow pattern whe
28. of the separator The purpose of this sensor was to estimate the air flow through the control valve Experimental testing and verification 15 3 2 Data Flow and Data logging The data measured by the devices installed on the Miniloop had to be recorded analyzed and stored To accomplish this the different devices are connected to a lap top computer through the Field Point modules Figure 3 11 The Field Point modules are mounted on a terminal base inside a water proof locker The different analogue transducers are connected to the FP input module middle while the control valve 15 connected to the FP output module right The computer is connected to the communication module to the left as DO H LI mea HE NATIONAL INSTRUMENTS AS 13283 Brant pat A D mm Figure 3 11 Picture of the FP modules mounted on the termination card 3 2 1 Software and drivers The hardware FP modules and software required to analyze store and display the data are delivered by National Instruments NI The software needed is LabView with the following additional content installed e PID control toolset Fieldpoint explorer version 3 01 drivers FP module drivers Experimental testing and verification 16 3 3 LabView LabVIEW delivers a powerful graphical development environment for signal acquisition measurement analysis and data presentation giving the flexibility of a programming language wi
29. position The slug sensor signal is send to this SubVi and the density is calculated based on the equations inside the SubVi This subVi needs the density calculated in the previous subVi the pressure drop across the valve and the valve opening It will then estimate the total mass flow through the valve This subVi will calculate the valve opening from the actuator position 4 2 Filters and charts The block diagram in figure 4 1 are marked with green blue and red circles They are there so the user can quickly identify the following components Green circles Blue circles Red circle The componets inside the green circle are indicators They plot the corresponding data in the charts located in the front panel The different measurements being plotted are downstream pressure P1 upstream pressure P2 valve position 7 flow Q and W and the slug sensor 52 The blue circles encompass the PID Control Input Filters These filters apply a fifth order low pass FIR filter to the input value Filter cut off frequency is designed to be 1 10 of the sample frequency of the input value Use this function to filter measured values such as process variable in control applications The red circle encompasses a lag filter This filter has been added to the estimated mass flow measurement The filter parameters can be adjusted in the filter box on the front panel 11 4 3 CASE structure and the controllers The different control structu
30. program The components of a block diagram are lower level VIs built in functions constants and program execution control structures You draw wires to connect the appropriate objects together to indicate the flow of data between them Front panel objects have corresponding terminals on the block diagram so that data can pass from the user to the program and back to the user The hierarchy window displays a graphical representation of the calling hierarchy for all VIs in memory including type definitions and global variables This hierarchy window shows the relationship between the subVIs in a program This is a good insight to the structure of the program The power of G programming lies in the hierarchical nature of VIs After creating a VI one can use it as subVI in the block diagram of a higher level VI Experimental testing and verification 17 3 3 1 Miniloop front panel Figure 3 13 shows the front panel for the LabVIEW program miniloop The original program created by B rdsen 3 were abandoned and a new one was created from scratch to accommodate better flexibility and different control structures The front panel serves as the interface between the user and the lab scale Miniloop 0 005722 Gu 29 997 qo fin 11 000 Tom STOP 10000 30 110 10 100 To MUSK 15 0000 SR 10 000 B ET dei joo dei 10 0 det 40 0 Gg 10 020 SE 4 50000 Jooo 3E 120 de j
31. rdi senere 4 TEACH s 0 Gr n PU uttI 3 Niveau displayet lyser r dt 4 S t mode switchen tilbage p RUN Et punkts Teach uden emne 1 S t mode switchen p SET 4 Tryk p TEACH knappen i ca 1 sekund uden emne SET WH 2 Tryk p TEACH knappen i ca 1 sekund TEACH lt lt 15 act lt lt 5 Teach er udf rt n r niveaudisplayet skifter fra r d til gr n Niveau displayet vil vise den aktuelle v rdi senere 3 Teach er udf rt n r niveau displayet skifter til r d Niveau displayet vil vise den aktuelle v rdi senere dy Gr n 4 S t mode switchen tilbage p RUN 6 S t mode switchen tilbage p RUN CM RUN CD RUN Note Om man udf rer Teach med emnet f rst eller emnet sidst 5 Indstillingsv rdien bliver automatisk indl rt n r n ste betyder intet emne passerer fiberen i ae P Pr cis positions Teach 1 S t mode switchen p SET SET EMI 1 2 Tryk p TEACH knappen i ca 1 sekund uden emne e Note Hvis et punkts teach ikke kan udf res er forskellen mellem emnet og baggrund for lille Pr v da to punkts teach TEACH TF 15 E3X DA N 3 Niveau displayet lyser r dt 4 Anbring emnet i den nskede position og tryk p TEACH knappen i min 3 sekunder TEACH TI lt 3s Dimensioner Note Alle enheder er angivet i millimeter hvis ikke andet angivet Forst rker Forst r
32. supply first valve V1 first then the air supply valve V2 Disconnect the power supply to the pump Turn of the field point modules with the switch Disconnect the power source to the field point modules Close LabView and shut down the computer ass Comments e The air supply must always be turned on first and shut down last The reason is to obtain a certain pressure inside the hose to prevent backflow of water into the buffer tank BT The pump will start to work as soon as the power supply is connected So make sure there is no air in the pipe leading from the water reservoir to the pump Also make sure that the water level in the reservoir tank 1s higher then the outlet leading to the pump e The valve will always close itself when the field point modules are turned off The miniloop program therefore has to be put in run mode to open the valve before air or water is introduced to the system Failure to do this will result in a quick pressure build up in the pipe and a blow out of the pressure sensors Adjusting flow rates During system start up it is recommended to adjust the air flow to the desired flow rate before introducing water into the system Once the air flow is adjusted the water flow rate can be adjusted to the desired rate Take note that the water flow rate will vary depending on the upstream pressure The water flow will normally vary around 10 To maintain consistency it s recommended to use the max flow rate d
33. the volt signal in figure 3 22 to a hold up measurement some algebraic code were added to Labview These take into account the geometry of the tube and scale the hold up to take values between 0 and 1 Had this been the case the slug sensor would probably not have been useable The surface of the bubbles would have deflected the light away from the optical sensor resulting in severe disturbances Experimental testing and verification 25 3 4 2 Estimating the flow through the choke valve The mass and volume flow through the choke valve can be estimated by measuring the pressure drop over the choke and the mixture density Multiphase flow through a choke valve is complex but according to Skogestad there have been successful implementations of a cascade controller in the industry by using a simple valve equation for liquid flow An attempt was therefore made to estimate the flow from equation 2 1 The mixture density could be estimated from the hold up measurement x by the following equation Pn X Pair 3 1 The biggest challenge lay in experimentally deciding the valve characteristics The only available measurements were the flow rates into the system The slugging nature of the system also made it impossible to get any experimental stationary open loop values when using both gas and liquid For that reason it was decided that the only viable option was to decide the valve characteristics using only liquid flow Whe
34. time s Figure A 5 Open loop data for z 0 3 Appendix A 50 25 Valve opening 0 16 gt 0 14 2 0 12 a o 0 08 0 06 0 20 40 60 80 100 120 time s 0 15 D S 0 1 N 0 05 0 DDD AVM 0 20 40 60 80 100 120 time s Figure A 6 Open loop data for z 0 25 22 Valve opening 0 16 3 0 14 8 0 12 a 04 0 08 0 06 0 20 40 60 80 100 120 time s 0 15 D g 0 1 N oO 0 05 0 0 20 40 60 80 100 120 time s Figure A 7 Open loop data for z 0 22 Appendix A 51 20 Valve opening 0 16 gt 0 14 3 0 12 o 0 08 0 06 0 20 40 60 80 100 120 time s 0 15 o S 0 1 N 0 05 0 0 20 40 60 80 100 120 time s Figure A 8 Open loop data for z 0 2 19 Valve opening 0 16 0 14 8 0 12 a 0 1 0 08 0 06 0 20 40 60 80 100 120 time s 0 15 D g 0 1 N al 0 05 0 0 20 40 60 80 100 120 time s Figure A 9 Open loop data for z 0 19 Appendix A 52 18 Valve opening 0 16 5 0 14 2 0 12 a 04 0 08 0 06 0 20 40 60 80 100 120 time s 0 15 o S 0 1 N 0 05 0 0 20 40 60 80 100 120 time s Figure A 10 Open loop data for z 0 19 16 Valve opening 0 16 3 0 14 8 0 12 a 0 1 0 08 0 06 0 20 40 60 80 100 120 time s 0 15 D g 0 1 N al 0 05 0 0 20 40 60 80 100 120 time s Figure A 11 Open loop data for z 0 18 Appendix A 53 14 Valve opening 0 16 5 0 14 3 0 12 a 0 1 0 08 0 06 0 20 40 60 80 100 120 time s 0 15 o S 0 1 N 0 05 0
35. upstream pressure P1 is actually very good for the entire valve opening range The fit for the downstream pressure 1s acceptable for the low to medium valve openings The deviations in amplitude for higher valve openings are acceptable in light of the priorities given Part of the deviations for the downstream pressure fit may result from the disturbances associated with the downstream measurement The downstream measurement 15 also located 15 cm below the valve witch will cause the experimentally measured pressure drop to be a bit higher then it should By examining the bifurcation diagram it can be seen that the experimental pressure drop over the choke is oscillating to a lesser degree in the stable regime The mayor cause for this behaviour is caused by the flow behaviour as described by figure 3 21 The frequencies of the oscillations are not included in the bifurcation diagram Figure 3 30 and 3 31 shows the open loop behaviour for a valve opening of 30 for both the model and the lab scale Miniloop The amplitude of oscillations for the upstream pressure P1 are almost the same for both the model and the Miniloop The model calculates a bit lower amplitudes for the downstream pressure P2 By examining the frequency of oscillations it can be seen that the slug frequency is about 10 higher for the Miniloop compared to the simplified model In this case the model was tuned to achieve a good fit for the amplitude Since the upstream gas volume is
36. white line represents the estimated mass flow through the top side choke The left side of the chart displays the Miniloop in open loop mode while active control 1s used on the right side The estimate of the mass flow could now be used for control purposes Experimental testing and verification 29 3 5 Open loop experimental data The Miniloop was run in open loop and the valve opening was gradually changed from fully open till fully closed The corresponding pressure drops were recorded and analyzed to create the bifurcation diagram figure 3 27 0 25 Maks trykk Min Trykk 0 2 P barg 0 1 0 05 n 0 10 20 30 40 50 60 70 80 90 100 0 08 0 06 0 04 NOR mu P barg 0 02 0 10 20 30 40 50 60 70 80 90 100 Ventil pning Figure 3 27 Bifurcation diagram for the experimental data As can be seen from figure 3 27 the system 15 stable up to a choke opening of 19 If the choke is opened further the system will enter the slug flow regime The system becomes unstable and the pressure will start to oscillate Early solutions to the slug problem in the oil industry utilized this property By increasing the pressure drop over the top side production choke the slugging was successfully eliminated However this solution also increased the total pressure drop in the well pipeline system resulting in lower oil reco
37. within the stable region Attempts to force the process into the unstable region resulted in the reappearance of slugging Simulations on the simplified slug model however managed to stabilize the system in the unstable area The reason the cascade controller failed to stabilize the Miniloop is because of the disturbances and the noise picture associated with the mass flow measurement In light of the control objective given in chapter 1 this cascade configuration did not achieve acceptable control but both the simple PI controller and cascade controller using the upstream pressure did The simulations done with the simplified model are in good accordance with the results obtained experimentally The response and behaviour of the simulations described the real process very well This thesis therefore adds to the growing list of papers that verifies the simplified slug model as a useful tool for control purposes References 46 References 1 Golan M Internal Notes on Orfice Valve Equation Thornhill Carver Equation NTNU 2003 2 Storkaas E Skogestad S low dimensional dynamic model of sever slugging for control design and analysis NTNU June 2003 3 Bardsen I Anti slug control for two phase flow Experimental verification In Norwegian NTNU autumn 2003 4 Storkaas E Skogestad S Stabilization of sever slugging based on a low dimensional nonlinear model NTNU 2002 5 Kaasa L Dem
38. 2 TC200F E32 TC200F4 670 1 mm 4 000 530 3 700 200 1 400 E3X DAI Fleksi bel fiber Meget hardfor Fleksi bel fiber Meget hardfor E3X DAI Ideel til montage pa korende dele Ideel til montage pa korende dele E3X DAI E3X DAL E3X DAL Lang tasteafstand pladsbesparende 3 mm dia E3X DAB11 N E3X DAI Til sma emner 1 dia 1 2 Note der overstiger v rdien 1000 p displayet Fibrene p E32 T17L er 10 m 1 E32 T21 4mm Ao E32 T22B E32 T14L E32 T24 St rrelsen pa standardobjektet er den samme som fiberens diameter anvendes linse er det linsediameteren Det mindst aftastbare emne med sender modtager er fundet nar fotoaftasteren i digital 12 bits visning er indstillet til at modtage lys Indikerer v rdier for standard mode E3X DA N Appli kation Modst r kemika lier og andre h rde milj er Lille spredni ng af lyskegle Af tastning af areal Af tastning af areal Note 1 2 2 Maks vedvarende temperatur er 130 C 3 ndikerer varmemodstanden spidsen Indikerer v rdier i standard mode Egenskaber Teflon belagt Omgivelsestemperatur imellem 30 C til 70 C Teflon belagt vinkelaftastning Omgivelsestemperatur imellem 30 C til 70 C Modstar 200 C fleksibel Teflon belagt 1 Omgivelsestempera
39. 26 Pino the 7 to tie datapOLDS Var P Snapshot of the chart showing displaying the measured mass flow 28 Bifurcation diagram for the experimental data 29 Model characteristics with important parameters 30 Bifurcation diagram for the simplfied slug 30 Open loop behavior for the miniloop Z 0 3 see 32 Open loop behaviour for the simplified model 2 0 3 32 Real partof the worst pole eoa UPS ete ti ve uos 34 Performance of a pressure controller on the miniloop 35 Performance of a pressure controller on the simplified model 36 Cascade control with W in inner loop and P1 in outer loop miniloop 38 Cascade control with W in inner loop and P1 in outer loop Modell 40 Cascade control with W in the inner loop and P2 in the outer loop 4 Cascade control with W in inner loop and P2 in outer loop model 42 Figure A I Figure A 2 Figure A 3 Figure A 4 Figure A 5 Figure A 6 Figure A 7 Figure A 8 Figure A 9 Figure A 10 Figure A 11 Figure A 12 Figure A 13 Figure A 14 Figure A 15 Open loop dala TOP V BE 49 Open loop dald TOEZ Var 48 Open loop data TOt z D cipe tata vtto rea e qa Da eb iN Sos ream EE ps 48 Open loop data
40. 3 8 Figure 3 9 Figure 3 10 Figure 3 11 Figure 3 12 Figure 3 13 Figure 3 14 Figure 3 15 Figure 3 16 Figure 3 17 Figure 3 18 Figure 3 19 Figure 3 20 Figure 3 21 Figure 3 22 Figure 3 23 Figure 3 24 Figure 3 25 Figure 3 26 Figure 3 27 Figure 3 28 Figure 3 29 Figure 3 30 Figure 3 31 Figure 3 32 Figure 3 33 Figure 3 34 Figure 3 35 Figure 3 36 Figure 3 37 Figure 3 38 How ee E are T 6 Plow Sheet For the MNOO HP 10 RATE TG TOR ve 11 Rate MEET US 11 IIS 12 SIF EE 12 Ps 12 Keen 13 13 he eg AT ALON NE pL RUM dere RE 13 CV ves 14 Picture of the FP modules mounted on the termination card 15 Labview FO Panel endete oo D E 16 The front panel for the labview program muniloop 17 The data flow inside the block diagram 19 DeCHon ot tle DIOC Ee TS 20 The content of the sub VI named calibrate 21 Section Of the hierarchy windOW e re Reb EEN 21 Initial readings from the slug sensors eee sees eerrrrrreeerreee 22 The slug sensor after the colouring matter was changed 23 The slue sensor after signal SAN 23 olus Tow pattern Ale PIPE vvs 24 Final She Sensor TEAGIN OS Larssen 24 Pressure and flow vs Valve Opening saa 25 OZ SN Pee ee
41. 35 mm 35 mm 25 mm 10 mm 25 mm 10 mm 25 mm 10 mm Det mindst aftastbare emne med sender modtager er fundet n r fotoaftasteren i digital 12 bits visning er indstillet til at modtage lys der overstiger v rdien 1000 p displayet 1 Teflon er et registreret varem rke 5 Fibren er 2 m E3X DA N OMRON E3X DA N Typer for diffus reflektion 4 Indikerer modeller hvor det tilladt at forkorte fiberen ved hj lp af medf lgende saks ME Super lang tasteafstand P Standard mode Super high speed mode Lang E3X DAL taste afstand E3X DAB11 N E3X DAL 3 mm dia E3X DAL E3X DAL E3X DAI E3X DAB11 N E3X DAL 5 ia 90 mm 40 mm E3X DAI B jeligt metal Ti fiberr r i enden M6 2 5 mm dia E Dou 1 2 mm dia 90 mm 40 mm ES3X DAI Bojeligt metal fiberror i enden M3 de E32 DC200F4 0 8 mm dia E3X DAL pr cis aftastning 0 5 mm dia 5 mm dia E3X DAL pr cis aftastning Kan b jes som en ledning B jningsradius 1 E3X DAI 44 tasteemne 500 x 500 0 01 mm dia 300 x 300 0 01 mm dia 200 x 200 0 01 mm dia 400 x 400 0 01 mm dia 100 x 100 0 01 mm dia 400 x 400 0 01 mm dia 100 x 100 0 01 mm dia 300 x 300 0 01 mm dia Applika Egenskaber Udse
42. Abstract The pipeline riser systems needed to transport oil to the production facilities gives rise to an undesired flow regime known as slug flow Slug flow usually occurs when you have a low point in the pipeline topography followed by an inclining section of pipe These slugs can grow very large and often cause severe problems when they reach the production facility A lab scale Miniloop had been build to simulate severe slugging by a previous student A simple PI controller using the downstream pressure as measurement stabilized the process and eliminated the slugging However this measurement can in many cases be unavailable A controllability analysis will show that the process can be stabilized by using a cascade configuration with a flow measurement in the inner loop and the upstream pressure in the outer loop This thesis will document the changes done to the Miniloop in order to obtain the measurements needed for the cascade controller To document this better a user manual has been written Two different cascade structures were tested on the Miniloop and simulated on the simplified slug model A cascade controller using mass flow in the inner loop and downstream pressure in the outer loop managed to stabilize the process and eliminate the slugging The cascade controller using only upstream measurement with mass flow in the inner loop and the upstream pressure in the outer loop failed to stabilize the Miniloop However it stabilized the s
43. J i L d IT L 1 0 1 2 3 4 5 6 7 8 9 10 time min 1 E 3 0 5 2 3 fe a dag Ur ry if d 0 0 1 2 3 4 5 6 7 8 9 10 time min Figure 3 35 Cascade control with W in inner loop and P1 in outer loop miniloop Experimental testing and verification 39 By examining the actuator usage one can see that the system 15 stabilized at a valve opening in the unstable region for open loop When the controller is turned on the slugging is quickly eliminated but the cascade controller uses a bit more time to reach the reference value compared to the pure PI controller in chapter 3 8 1 This is outweighed by the better tracking performance exhibited by the cascade controller The amplitude of the pressure oscillations during active control is small meaning that the controller successfully keeps the pressure tighter around the reference value compared to the PI controller This didn t come as a surprise Because of the low gain in the inner loop the actual stabilizing 15 done by the outer loop In the case above the inner loop merely serves as a filter for the outer loop Tuning the controllers turned out to be a difficult and time consuming task of trial and error An attempt was made at disconnecting the outer loop and tuning the inner loop first When this proved unsuccessful the outer loop was reconnected and both loops were tuned simultaneously The overall strategy of the tuning was as follows By operating at
44. Q 0 1673 0 8720 0 63471 0 0958 13 027 5 6430 0 8720 0 63471 0 0226 0 0092 0 06141 0 1530 0 0092 0 06141 0 0401 Experimental testing and verification 34 The upstream pressure P has one LHP zero Since LHP zeros imposes no fundamental control problems P would be the obvious choice of measurement However this measurement can in many cases be either unreliable or unavailable and other measurements have to be considered From the bandwidth limitations imposed by equation 3 6 there cant be any RHP zeros smaller then 0 7605 Of the alternatives in table 3 3 pm have RHP zeros close to the unstable poles This makes it unsuitable as a measurement for a stabilizing controller due to the bandwidth limitations imposed From table A 2 it can be seen that the zeros for increases as the valve opening increases At the operating point of z 0 3 it is larger then 0 7605 and cant be directly dismissed as a possible measurement The model in 4 gets lower zeros for P conludes that it cant be used for a stabilizing controller Both Volume flow and mass flow W appears to be better alternatives but they both have LHP zeros close to or at the imaginary axis An attempt to stabilize the system with one of these measurements would result in an almost integrating closed loop system According to Storkaas a cascade control could solve this problem by using a combination of a flow measurement and some other measurement e g pressure 0 6 UM
45. The colouring matter used to give the water a blue colour is called Vulcanosol Blau 684 More colouring matter can be obtained from Engineer Arne Fossum at his office in K3 019 Very little substance is needed to dye the water It s recommended to gradually add small amounts of substance until the desired slug sensor value has been achieved The system should be set to pump water through the system to disperse to substance properly One spatula is enough to dye all the water in the system if the reservoir tank 1s half full 17 6 References 1 B rdsen I Anti slug control for two phase flow Experimental verification In Norwegian NTNU autumn 2003 2 S ndrol M Anti slug control Experimental testing and verification NTNU spring 2005 18 Appendix Equipment suppliers and prises Table A 1 lists the suppliers and prises for the different components in the Miniloop Table A 1 List of equipment suppliers and prises NOK Rate meter for water Gemu 3021 J S Cock 3991 3991 Control Valve Gemu 554 P O BOX 68 Stovner 4502 N 0913 OSLO Phone 47 22 21 51 00 D 5110 HAB Flow Teknikk as 9914 Rate meter for air Olav Brunborgsv 27 P O BOX 244 1377 Billingstad Phone 47 66 77 54 00 A1 Module FP AI 100 National Instruments 2745 2745 AO Module FP AO 210 P O BOX 177 3555 Termination base FP TB 2 N 1386 Asker 1512 Communication FP 1000 Phone 47 66 90 76 60 3105 module Signal transducer for MICROANALOG JF Kn
46. al version _ seite 10 Readout systems seite 11 Contact addresses Seite 12 SIINTILE LEN IINS TT Basic structure of the Ass STREAM amp flow sensor M ASS STREAM Features and Applications sm VVorth knowing ass STREAM is the synonym for a metering principle having the following advantages Smallest standard range 0 005 0 1 I min Air Highest standard range 100 0 7500 0 min Air Lower and higher ranges available on request am Features Low pressure drop Installable in virtually any position Rugged design No moving parts Lower sensivity concerning dirt and humidity Bodies in stainless steel or as a more economical Measuring independent of pressure and temperature changes aluminium version zm Applications Gas consumption metering Exhaust gas metering Semiconductor industry Analytical intruments N 0 generators Fuel cells Mechanical engineering And much more zm pt ONS Low AP version Integrated totalisation Integrated actual display Integrated setpoint potmeter Readout systems M ass Flow Meters Principle of Operation Flowmeters of M W Instruments are sul table for all kinds of applications in indu strial chemical medical and laboratory environments Usuable for virtually every kind of gases No moving parts Very l
47. ay close attention to the system if these values are changed Wrong parameters during active control can make the valve close it self leading to a pressure build up and a blow out of the pressure sensor If this happens the user must switch the control selector back to no control to reset the valve position and prevent the build up of pressure PID CONTROL BOX Control Selector No control Parameters for single loop control with P1 as measurement Sp baro Jono cc 10 100 co BE 15 0000 DEI USE ooo vi oo oo Tals joo Parameters for kaskade control with Parameters for kaskade control with vw in inner loop and P2 in outer loop Quter loop disconnected Inner loop Sp barg Kp Ti s Sp og Kp bar Til IST g t IT 1 110 Figure 3 2 PID control box 3 4 Manual control a When the control selector is set to No control in the PID control box the process will run in open loop mode The user can adjust the valve opening by adjusting the slide bar or by entering the new value for the valve opening in the small box below Figure 3 3 Manual valve control 4 The Miniloop block diagram In this chapter the most important components or subVi s in the block diagram will be briefly explained Understanding of the block diagram is essential if the user wish to add more code or alter the existing co
48. clude modification of existing equipment and purchase of new parts In the original loop permanganate was used to dye the water red The Miniloop had been out of use for some months so the colouring matter used had stained the pipes These stains created problems for the optical sensors For reasons that will be made obvious in chapter 3 4 1 all the tubes were replaced and a more water soluble colouring matter was added The new colouring matter added is called Vulcanosol Blau 684 The original brackets used to attach the slug sensors to the pipe had a couple of flaws In the original bracket there was no way to adjust the distance between the two optical cables There was also some concern that the metal used in the brackets could reflect some of the light and create an error in the measurement After some consideration a new design was chosen The new bracket figure 3 5 was drilled out of a PVC pipe to resemble a horseshoe With this design the distance between the two cables could be altered depending on how far into the material the cables were screwed One of the biggest problems with the original loop was that we had no way of ensuring the same operating conditions each time an experiment was conducted There was a flow meter installed to measure the flow of water but we lacked a way to measure the flow of air For this reason a flow meter for air was purchased and installed An additional pressure sensor was also installed at the air outlet
49. d Driftm de Tidssekvensdiagram Udgangs Udgangskredsl b gang v lger status NPN E3X DA11 N E3X DAB11 N E3X DA6 E3X DA14V E3X DA21 N E3X DA7 PNP E3X DA41 N E3X DA8 E3X DA44V E3X DA51 N E3X DA9 LIGHT ON L ON DARK ON D ON LIGHT ON L ON DARK ON D ON LIGHT ON L ON DARK ON D ON LIGHT ON L ON DARK ON D ON Haendelse ON H ndelse OFF Udgangsindikator ON orange Udgangs transistor OFF ON OFF Belastning ON rel OFF H ndelse ON Haendelse OFF mellem brun og sort Udgangsindikator ON orange Udgangs transistor Belastning rel Haendelse ON H ndelse OFF OFF ON OFF ON OFF mellem brun og sort Udgangsindikator ON orange Udgangs transistor Belastning rel H ndelse ON H ndelse OFF OFF ON ON OFF mellem brun og sort Udgangsindikator ON orange Udgangs transistor Belastning rel Haendelse ON H ndelse OFF OFF ON OFF ON OFF mellem brun og sort Udgangsindikator ON orange Udgangs transistor ON Belastning ON rel Haendelse ON H ndelse OFF OFF mellem brun og sort Udgangsindikator ON orange Udgangs transistor Belastning rel Haendelse ON H ndelse OFF OFF ON OFF ON OFF mellem brun og sort Udgangsindikator ON orange Udgangs transistor ON Belastning ON rel Haendelse
50. de More information about the detailed tasks of each component can be learned by reading the text boxes inside each subVi or by using the help function in LabVIEW 4 1 The subVIi s oe preassure 1 Em ELL Hr Upstream preas Ste ia E IUUD UU UU LS DE Let Tet Le cL Le kk nn 2 Valve pasition Array ues 2 Figure 4 1 The block diagram Figure 4 1 shows a section of the block diagram The yellow boxes with name tags in the middle are different SubVi s Each of them contains additional code that can be accessed by double clicking the block in the actual program The programming follows the flow of data A while loop encompasses most of the code and the measurements will enter the while loop in the wire marked with a big red X Once it 1s inside the while loop it will pass from box to box by following the different wiring The different subVI s and their purpose are listed below Read Here the data 15 indexed and tagged Calibrate The data is split into separate data streams and calibrated Value check Here the different measurements are checked to ensure they don t take inconsistent values Disturbance and noise may cause a measurement to take illegal values 1 e a pressure becoming negative This subVi will remove these values and force the measurement to take values within a given limit 10 Density Flow Valve
51. e 2 1 The slug flow cycle Theory 7 2 2 Flow through chokes Knowledge about the behaviour of flow through chokes 15 important in production systems where flow rates are controlled by choke valves Different phases like gas water and oil have different flow behaviour through choke valves and other restrictions 2 2 1 Single phase liquid flow through chokes The flow rate through a valve depends on the size of the valve the pressure drop over the valve and the fluids properties according to the following equation 11 AP C z 2 1 p Where Qi Liquid flow rate l min C Valve constant Z Valve opening 0 z lt I AP Pressure drop over the valve p Density of fluid 2 2 2 Gas Flow through chokes Gas flow through chokes is more complex then fluids because of its compressibility and pressure temperature changes Therefore corrections have to be made to the equation for expansion and temperature A correlation for the gas choke is given by 1 _ TECUM Q 76 85 10 7 icd 2 2 Where Gas rate standard conditions Sms s Cd Discharge coefficient A2 Choke area me Z1 Z factor at choke inlet T Temperature at inlet K ag Specific gas gravity relative to air Di Inlet pressure Pa k Adiabatic gas constant y Expansion ratio Theory 8 2 2 3 Multiphase flow through choke When two or more phases flow together in a pipe many different flow regimes may occur T
52. e two tubes connected to the tank The top tube is connected to the separator and the water will return to the reservoir through this tube The second tube is connected on the flat end side of the tank This is connected to the pump witch pumps the water into the Miniloop Before the pump is turned on the user has to make sure the liquid level is higher then this outlet if not air may enter the tube This can damage the pump and in the worst case damage it The water level should at least be 3 cm higher then the outlet Additional water can be added to the tank through the open hole on the flat side above the outlet There is a water source available with a long enough hose in close proximity to the loop To empty the tank the user can remove both tubes and pour the liquid into the sink If the user wish to clean the tank with water make sure it is not to hot Using only hot water to clean the tank may cause the glass to crack jt UR 3 d 1 i Figure 5 1 Reservoir tank 5 2 Buffer tank The buffer tank 1s used to create enough upstream volume for the gas This is a prerequisite for slugging to occur The volume available for the gas will influence the amplitude of the pressure oscillations The volume can be altered by adjusting the liquid volume inside the tank To add more water the user has to disconnect both tubes leading to it As for the reservoir tank the buffer tank mustn t be cleaned with to hot water as this may resul
53. een the slugs Slug flow can be divided into two main types hydrodynamic and gravity induced slugging Hydrodynamic slugging occurs in horizontal pipelines because of velocity differences between the phases and will not be a topic in this thesis 2 1 1 Gravity induced slugging Gravity induced slug flow 15 induced by a low point in the pipeline topography followed by an inclining section of the pipe The prerequisite for this to occur are low pipeline pressure and flow rates A sufficiently large volume upstream of the slug is also needed to allow the build up of gas The slug cycle can be divided into four stages as shown by figure 2 1 It is initiated by an accumulation of liquid in the low point of the pipe stage 1 This will eventually block the flow of gas and lead to a build up of pressure upstream the liquid slug The pressure and liquid slug will continue to grow stage 2 until the pressure is high enough to overcome the weight of the liquid in the riser At this point the gas will start to penetrate the liquid and push the slug all the way out of the riser stage 3 This will result in a drop of pressure and the gas will no longer be able to drag the liquid up the raiser Some of the liquid will therefore fall back down the riser stage 4 and accumulate at the low point and initiate the cycle again re n a Deg ml nuam T m EA T Mace i Sege wm Oe Ema E Figur
54. elereliCeS uer eege eege 18 Appendix A Equipment suppliers and 56 5 19 Appendix B Ed ipment EE 20 1 Introduction The Miniloop was originally constructed by B rdsen 1 as a part of his fifth grade project with the Department of Chemical Engineering at NTNU Since then some work has been done on the Miniloop by S ndrol as a part of his thesis New measurements have been added and analyzed new user interface has been constructed to obtain the new measurements and to allow different control structures This user manual was written as a part of the thesis 2 however it is meant to be a stand alone user manual This means that some of the things presented in this manual can also be found in the Diploma thesis 2 The Miniloop is essentially very easy to use However there are some issues the user should be aware of It is therefore recommended to read this user manual before performing any experiments on the Miniloop 2 Miniloop and equipment Figure 2 1 shows an overview of the lab scale Miniloop The different components are listed in table 2 1 FF modules E nou EI n r PL Figure 2 1 Flow sheet for the Miniloop As can be seen from the figure the Miniloop has a water WT sorce and an air source The water 15 pumped from the reservoir into the system while the air is let into the system from a pressurized air outlet in the wall The flow rate of water and air is controlled by manually adjus
55. ende Forsteerker Tasteafstand mm 1 Standard Tilladelig tion objekt b jnings min radius 2 Guld tr d E32 D11 LP E32 D12 E32 D21L E32 Ab off E32 E E32 DC200E 2 200 E32 DC200B4 E E32 DC200F E32 DC200F4 E32 D11R 7 1 mm E32 D21R 25 25 mm 25 mm E3X DA N Flek Ideel til montage p sibel fi k rende dele ber Meget hardfor E3X DAL E3X DAL Note 1 V rdien er opgivet for et standardobjekt 2 Det kan v re n dvendigt at udf re et eller to punkts teach for at opn samme resultat 1 Tasteafstand ved hvidt papir 2 indikerer v rdier i standard mode Egenskaber Udseende Forst rker Coaxial M6 coaxial E3X DAL fiber Stor aftastnings n jagtighed ved lang afstand 3 mm dia coaxial E3X DAL Stor aftastnings n jagtighed 3 mm dia M3 coaxial Stor E3X DA aftastnings n jagtighed muligt at montere linse for smalt spot E39 F3A 5 F3B F3C M3 coaxial Stor E3X DAL aftastnings n jagtighed muligt at montere linse for smalt spot E39 F3A 5 F3B F3C 2 mm dia coaxial ESX DAL Stor aftastnings n jagtighed muligt at ee linse tor ammalia smalt spot 0 1 til 0 6 dia E39 F3A 2 mm dia coaxial Stor aftastnings n jagtighed muligt at montere linse for smalt spot 0 5 til 1 dia E39 F3A Vinkel aftastning 6 mm dia lang tastea
56. enings above 30 d f z VAP for0 lt 0 3 3 5 f 0 3 VAP forz gt 0 3 Estimating the volume liquid flow through the choke was now possible Unfortunately it proved impossible to get an estimate of the volume gas flow through the choke with the current measurement setup This in turn made it impossible to estimate the total volume flow Q The total mass flow W was still a viable option though The mass flow of gas was assumed to be much smaller then the mass flow of the liquid So setting the total mass flow W equal to the mass flow of liquid would not introduce to big an error A simple rearrangement of equation 3 5 would therefore give an estimate of the total mass flow Experimental testing and verification 28 SPA AP o for 0 0 3 3 6 f 0 3 4AP o for z gt 0 3 Where n is a tuning parameter set to 1 3 The system was forced into steady state by using active control A simple mass balance consideration now implied that the total mass flow through the choke had to be the same as the mass flow into the system The parameter n could then by tuned so that the estimated mass flow W would be the same as the measured liquid flow into the system LI b iJ ON WW Figure 3 26 Snapshot of the chart showing displaying the measured mass flow Pasi d d _ L2 Figure 3 26 shows a snapshot of the flow chart on the miniloop front panel The red line is the measured flow into the system while the
57. er control configurations Because an experimental approach was chosen the different controllers were tuned experimentally The controller would be tuned until satisfactory control was achieved In this case satisfactory control meant that the slugging 1s eliminated by stabilizing the system at a valve opening that would normally result 1n severe slugging for open loop During the work on this thesis great emphasis was put on the fact that a third party should be able to continue the work with as little effort as possible The miniloop program was constructed so that it should be easy to use for a third party Text boxes etc were added to the code to explain the function of the most important components This was also emphasized in this report by including a detailed description of the work done A user manual was also written for the Miniloop 1 4 Notation The word miniloop will be used frequently in this thesis However it will be used in two different contexts Miniloop written with a Capital M refers to the lab scale Miniloop while miniloop spelled with a small m refers to the LabVIEW program written to control it Hence Miniloop 15 the physical equipment setup and the miniloop program is the user interface Theory 6 2 Theory 2 1 Slug flow Slug flow is characterized by intermittent axial distribution of liquid and gas The bulk of the liquid is transported as slugs of oil and water while the gas is transported as bubbles in betw
58. er of the front panel the user will find some additional indicators that displays additional information about the system These include density actuator position and digital displays for the flow When the system is running in open loop mode no control the valve opening can be set through the slide bar Most measurements are already filtered to some degree but since the estimated flow measurement 15 the one most prone to disturbances an additional lag filter has been added The parameters for this filter can be altered by changing the values in the filter box When the program is shut down it 1s recommended to use the big red stop button located on the front panel to ensure a controlled termination of the program including the writing of data from memory to hard drive The data will be stored in a text file with the following format Table 3 2 Format of stored data t msek S V Qinlet Westimated Z l min kg min All kinds of data can be written to the file including other measurement calculations etc Adding another source of data to the stored file is a simple matter All the user has to do 1s connect the measurement to be stored to a subVi in the block diagram Consult the user manual in appendix B to learn how Experimental testing and verification 19 3 3 2 Miniloop block diagram As mentioned the block diagram contains the source code for the program The diagram 1s too large to be dis
59. et and two outlets The air is released to the surroundings through an open hole in the top while the water is returned to the reservoir The buffer tank figure 2 8 is a cylindrical container made of transparent glass For slugs to appear the system needs a sufficiently large air volume The air volume in the tank can be altered by adding water to the tank The control valve figure 2 9 is located at the top of the riser before the separator inlet The valve requires a 24V power supply and is controlled by a signal to the actuator between 4 20 mA The relationship between the valve s actuator and the valve opening is linear To operate the actuator an external pressurized air source of 4 8 bar is required to counteract the spring power The lab has its own pressurized air source which was used for this purpose Figure 2 1 Flow meter Figure 2 4 Slug sensor Figure 2 7 Separator Figure 2 8 Buffer tank Figure 2 9 control valve 3 Operating the Miniloop 3 1 Start up and shut down procedures Start up 1 Start the computer and open the LabVIEW program miniloop Make sure valve V1 and V2 are closed Connect the power to the field point modules Turn the field point modules on by using the switch Connect the power to the pump Put the miniloop program into run mode Turn valve V2 until the desired air flow is reached Turn valve V1 until the desired water flow is reached PR SA S N Shut down Shut of the water
60. fstand 2 mm dia sm byggem l E3X DAL 2 E3X DAL 2 Tasteafstand 1 300 x 300 0 01 mm dia 100 x 100 0 01 mm dia Standard objekt min tasteemne 5 Guld tr d 500 x 500 0 01 mm dia 300 x 300 0 01 mm dia 100 x 100 0 01 mm dia 200 x 200 0 01 mm dia E3X DA N E32 D11 4 mm E32 D21 E32 D21 ff Tilladelig b jnings radius E32 CC200 SL E32 D32L Ao E32 C31 Og E32 C41 E32 C42 E32 D32 E32 D14L 25 mm E32 D24 OMRON E3X DA N Applikation Egenskaber Forst rker Tasteafstand mm 1 Modst r kemikalier og andre h rde milj er Teflon belagt 3 Om givelsestemperatur imellem 30 C til 70 C E3X DAI Modst r 150 2 Omgivelsestempe ratur imellem 40 C til 150 C Varme bestandig E3X DAL M6 Modstar 300 C 4 Tf n med rustfri spiral fib erbeskyttelse Omgi velsestemperatur imellem 40 C til 300 C 1 V rdien er opgivet for et standardobjekt E3X DAL Note 2 Det kan v re n dvendigt at udf re et eller to punkts teach for at opn samme resultat 1 Tasteafstand indikerer v rdier opn et med hvidt papir s 2 Max vedvarende temperatur er 130 C Wa 3 Teflon er et registreret varem rke f
61. ght that returns A signal of 1 volt means that no light has returned to the sensor while 5 volts means all the light has returned The original range of application of the optical sensors is as a precision sensing device The precise location of an object could be determined because the object would block part of the light beam hence reducing the amount of light transferred between the cables As long as only air was present 1n the pipe all the light would pass through the pipe and return to the sensor via the second cable The goal was to estimate the hold up of liquid by measuring the amount of light absorbed by the liquid phase Since the sensor was intended to be used on solid objects this turned out to be a big challenge As can be seen from figure 3 18 the optical sensors delivered a constant signal of 5 volt independent of whether it was liquid or air in the pipe After further experimentation it was Experimental testing and verification 23 discovered that the spiked were caused by a phase transition between air and liquid and vice versa When the light hit such a transition the angle of the liquid surface would deflect the light away resulting in the spikes witch indicated that no light returned to the sensor This still did not explain why the sensors showed the same value for both water and air There were some speculations that the water didn t absorb enough light for the sensor to measure a difference The solution proved to be as
62. he different phases will also exhibit different behaviour when passing through a valve Correlations need to predict both critical and non critical flow for all phases and several assumptions needs to be done Because of this complexity valve sizing and characteristics are usually based on experimental results 2 3 Modelling Storkaas 2 has developed a simplified dynamic model of multiphase flow for systems where severe slugging occurs The model covers both the stable limit cycle known as slug flow and even more importantly the unstable but preferred stationary slug regime This makes it suitable for controller design The model focuses on describing the observed macro scale behaviour rather then the detailed physics that governs the flow For more details about the model see Storkaas 2 The macro scale behaviour described by the model is e The stability of the solutions and the operational conditions as a function of choke valve opening e The nature of the transition to instability An unstable stationary solution at the same choke valve opening as those corresponding to severe slugging e The amplitude frequency of the oscillations Storkaas model is based on the following assumptions e Constant liquid level in the feed pipe witch implies o Constant upstream gas volume e Only one liquid control volume Two gas control volumes separated by the low point and connected through a pressure flow relationship Ideal gas beha
63. he pump Figure 3 6 Pump Experimental testing and verification The reservoir Figure 3 7 is a cylindrical container made of transparent glass It serves as the water source for the Miniloop and the water is returned to the tank from the separator The buffer tank Figure 3 8 1s a cylindrical container made of transparent glass For slugs to appear the system needs a sufficiently large air volume The air volume in the tank can be altered by adding water to the tank The separator 15 also a cylindrical glass container with one inlet and two outlets The air is released to the surroundings through an open hole in the top while the water is returned to the reservoir Figure 3 7 Figure 3 8 Figure 3 9 Reservoir Buffer tank Separator Experimental testing and verification 14 The control valve 1s located at the top of the riser before the separator inlet The valve requires a 24V power supply and is controlled by a signal to the actuator between 4 20 mA The relationship between the valve s actuator an the valve opening 15 linear To operate the actuator an external pressurized air source of 4 8 bar 15 required to counteract the spring power The lab has its own pressurized air source which was used for this purpose Figure 3 10 Control Valve 3 1 2 Changes to the Miniloop As mentioned above some changes have been made to the Miniloop since it was constructed by Bardsen These changes in
64. he rate of water in l min It provides a signal between 4 20mA depending on the rate of flow which 15 send to the computer The rate meter for air figure 2 2 is placed in front of the mixing point of water and air It has a digital display that shows the rate of air in percent of its operating area witch 15 0 2 2 l min The rate meter also provides a signal between 0 5 V which is send to the computer The pressure sensors figure 2 3 are one of Motorola s differential pressure sensors that delivers a signal between 0 2 4 5 V The relationship between voltage and pressure is linear and its operating area is between 0 100 The slug sensors figure 2 4 are fibre optical sensors Each slug sensor is made up of two fibre optical cables connected to a sensor The light emitted from the senor will travel out through one of the cables and back through the other The device will provide a signal between 1 5 V depending on how much light is transmitted between the two cables The pump figure 2 5 used is a standard aquarium pump It can deliver up to 38 l min and work against a head of 3 1 m Special care must be taken to make sure it doesn t pump air as this can damage the pump The reservoir figure 2 6 is a cylindrical container made of transparent glass It serves as the water source for the Miniloop and the water is returned to the tank from the separator The separator figure 2 7 is also a cylindrical glass container with one inl
65. her current junctions on application WM odel D 5121 MFC with LCD display L n EI _ Display optional T G with display T odel 0 6251 Display optional G j H B n 8 8 a D z 1 Model A B C DF GH I All specifications subject change without notice zum Conversion factors The ass STREAM amp Series flow meters are normally calibrated on air For use on other gases than air a conversion factor must be applied This factor is determined by applying a complex formula However for a number of common gases you will find the values below zum Conversion factor table 1 1013mbar und 0 C air temperature penes GaS me orm DIZ D 51xx Series GaS cU DO D 51xx Hr LH 0 15 1 01 tee en 200 140 CH 0 67 0 76 ee 0 99 SE e O Tre S OO mS oo 060 uUEM c E naar N50 ER SCH os T OE a eo TE a EE er Gan MM UM E COMM 608 138 en Sg EE IT Best accuracy 15 reached by calibrating the instruments und actual When using our D 62xx serie with balloon gases like helium and proce
66. her rate of oil recovery There are currently several successful implementations of control systems that stabilize the system under conditions that would normally lead to slug flow They are briefly discussed in Storkaas 4 and are all based on experiments and rigorous simulators like OLGA In Storkaas 2 4 the need for a simpler model is made evident and a simplified linear slug model with only three states is developed Introduction D B rdsen 3 constructed a lab scale Miniloop to experimentally verify the simplified slug model The Miniloop successfully verified the simplified slug model as a useful tool for analysis and control purposes A simple pressure controller using the upstream pressure as measurement managed to stabilize the flow This measurement can in many cases be unreliable or unavailable so the use of alternative measurements will be explored in this thesis 1 3 Scope of the thesis This thesis can be considered as a continuation of the work done by B rdsen 3 The main part of the thesis was experimental work and the overall goal was to stabilize the Miniloop using alternative measurements as proposed Storkaas 4 This meant that a lot of additional work had to be done on the Miniloop Additional equipment had to be bought and installed The measurements and sensors needed to be analyzed and adjusted The miniloop program user interface also had to be rewritten from scratch to obtain the new measurements and to allow oth
67. ich is a low point connected to an inclined section of the pipe At the top of the riser the multiphase flow passes the control valve before it enters the separator At this point the air is released out of the system through an open hole in the separator while the water is returned to the reservoir To monitor the behaviour of the system a combination of pressure measurements and slug sensors are used As can be seen from Figure 3 1 the water has a blue colour This is a necessity and not a just a cosmetic issue The reason Is that the slug sensors are optical in nature and the water had to be a colour that allowed it to absorb the light emitted from the sensors This will be addressed more thoroughly in chapter 3 4 1 Experimental testing and verification 11 3 1 1 Equipment Table 3 1 lists the different equipment used in the Miniloop Consult table A 1 in the user manual appended to this thesis for more info about the distributors and prises More detailed information about the different equipment can be found in the user manual appendix B Table 3 1List of equipment Notation Equipment FT W Rate meter for water Gemu 3021 FT A Rate meter for Air P1 Pressure sensor MPX5100DP Feed inlet P2 Pressure sensor MPX5100DP Valve P3 Pressure sensor MPX5100DP Separator air outlet SI Slug sensor E3X DA N 52 Slug sensor E3X DA N PU Pump Eheim 1060 WT Reservoir BT Buffer tank ST Separator CV Control valve The rate meter for water Fig
68. ioi height 5 55 mm Working conditions Ambient temperature 10 480 C Storage temperature 20 60 Type of medium liquid 120 120950 Medium temperature Ref 1 FVG U 10 480 Ref no 20 PYDF 10 80 C Working pressure 10 bar 20 C Gh eristic saa Technical Information on Plastic Materials General information Housing protection class to EN 60529 IP 65 Weight DN 25 600 g DN 50 1500 g Dimensions L x WX H seedimensional drawing Mounting position optional Mounting note Inlat autlat distances 5 x DN Mota We recommend installation of a dirt filter for filtering articles contained in the medium ash width 100 um Directives EC Machine directiva 98 37 EC EMC 89 336 EEC Measuring certificate Calibration data for water 20 C are included Materials Medium wetted parts Inner turbine components PVDF Body or FYDF SG shaft Glass ceramics 41203 Flow transducer Housing ABS Housing cover of messuring instrument size B PMMA Housing seal NER Housing bolt 1 4303 KH SC housing P amp R bolt V 36 2 4 8 Profile packing Nitrile rubber Further housing materials upon request Order specifications Nominal size Ref no Seal material ON 25 25 FPM Viton 4 DN 50 50 Body configuration Ref no Display position Straight through sea diagrams on last page Connection Ref no Functional profile Unions metric 7 T
69. ir TT mn yn A VE 20 0 4000 mm Air FEE D 6270 55 A A A 25 0 5000 Air 50640000 26 1000 20000 l min Air Lt dE ENTENTE ET 150 0 30000 l min Air E EU A m To META Er EGEN DEEP OE 300 0 60000 l min Air D 6280 116 Gl G1 70 141 143 35 FT er All specifications subject Higher flows and other current junctions on application to change without notice M ass Flow Controllers M FC mum Principle of Operation Based on the concepts of our flow meters compact flow controllers are also availa ble The modular construction solenoid valve is integrated to the base when flows are up to approx 500l min N equivalent mmm Features Suitable for almost all gases and mixtures No moving parts in the sensor Good response times Pl control loop No inlet pipe necessary D 62xx series zm Standard Flow Capacities Mass Flow Controllers M odel AAA BB AA 12 A When higher flows are needed an external valve is employed The following kv values are available 6 6E 2 0 35 1 0 for higher values please contact factory Optional integrated actual flow display Optional integrated totalisation display No maintainance required Almost independent to attitude Flow capacitiy Air intermediate ranges available l min Air Higher flows and ot
70. ir equivalent flow i e divide the desired flow by the conversion factor Only for controllers check if the pressure differential across the valve AP is within the limits Only for controllers check if the calculated Kv value is within the specifications allowed Technical Specifications mum easurement System Accuracy based on air calibration 3 FS incl non linearity better one s on request Repeatability mn 0 5 FS better one s on request Time constant 63 2 mm T 0 7 sec standard better one s s p 10 _ Pressure sensitivity mm 0 2 bar typical ai Attitude sensitivity d 010 Cd leakintegrity 0 lt 2x10 mbarlepe bredte K aaaaaaaaaa Range ll 3 1000 K jy aaa gases All gases compatible with materials chosen mm Temperature d FI Pressure 10 bar higher on application Warming uptime 30 min for optimum accuracy kfk RENEE RENEE within 30 sec for accuracy 490 FS mme Installation series D 5100 straight pipe upstream Series D 6200 Unrestricted zum Mechanical Part SENSON ss ALL LO Body ALSE 316L or Aluminium anodised please specify sm Sieves 5 Stainless steel Protection sms IPAQ IP 65 on application sms Supply voltage d 24 Vdc 10 or 15 Current peak values em Series D 5100 EE 15 mA max H
71. ived from the optical sensors The PID control is located at the lower left corner of the screen This is where the user chooses witch control structure to use The loop is set to no control by default but by clicking it you can choose from the available control configurations from a pull down menu The tuning parameters for the different controllers are also located here witch means the user can change them by simply entering the new value In the upper left corner of the front panel the user will find some additional indicators that displays additional information about the system Most measurements are already filtered to some degree but since the estimated flow measurement is the one most prone to disturbances an additional lag filter has been added The parameters for this filter can be altered by changing the values in the filter box 3 3 Active control To apply active control the user have to open the pull down menu in the PID control box and choose witch control configuration to use The control selector is set to No control by default Choosing a different control structure then this will immediately switch the system from open loop manual control to closed loop active control The chosen controller will use the relevant tuning parameters given in the PID control box The default parameters will stabilize the system at the given set point Both parameters and set points can be altered during active control However one must p
72. ix A 56 A 4 Hold up measurement The hold up measurement is calculated from the optical sensor slug sensor The slug sensor provides a signal between 1 and 5 volt where I volt means that the tube 15 filled with water and 5 volt means the tube 15 filled with arr The following calculations are added into the LabVIEW code and represents how the program calculates the hold up based on the signal received from the optical sensor The calculations 1s carried out within the subVi called density A linear relation 1s assumed between the amount of light absorbed by the liquid and the amount of liquid the light from the sensor has to pass through Equation A 1 calculates the height of the liquid h in the pipe based on the sensor signal S h 1 9 1 9 3 9 S 1 9 1 1 3 9 A 1 Equation A 1 is merely a scaling of the signal from 1 5 volts to 0 1 9 cm 1 9 cm is the diameter of the pipe Equation A 2 will calculate the area of the segment of pipe covered with water Consider a circle of radius a cross section of pipe and imagine that the liquid has reached height h measured from the lowest point on the circle Note that 0 lt h lt 2a The area A of the segment of the circle covered by the liquid is then given by a 0 95 2 A LT 0 932 wel AS 0 95 h h 2 0 95 h A 2 By dividing the area of the segment of pipe covered with water with the total area of the cross section we get the hold up A w a 095 vr
73. izer flow counting or batch controller dosing function gt Extremely low pressure loss Advantages gt Easy operation via keypad gt Process adaptable Short inlet and outlet distances gt Freely scalable measurement range Integrated flow rectifier Dimensional drawing GEMU 3021 mm 123 40 ER 50 61 1 2 140 50 187 63 63 103 i32 3 4 188 sz E ML 3021 Technical specifications Working medium Any liquid inert or corrosive aqueous media subject to the correct choice of body and seal materials Max perm temperature of working medium sea datasheet Technical Information on Plastic Materials Electrical specifications Power supply 18 30 V DC Power consumption P typ 1 W Currant consumption I for currant se Output signal Pulse output PNP LU typ Us 17 Vat24V 5 2 5 Vat 24 V 10 5 0 Vat 24 V 20 Pulsa rata K factor adjustable K factor see measuring certificatei Current 04 20 m Resolution 23 pA 10 1 5 bi Accuracy mere impedance 500 Ohm load control 0 25 Rel ei resistiva load tage 250 V AG 7 220 V Current 2 AAC 7 ADC Power BO W Electrical connection Plug according to DIM 43550 Gross saction of cabla 8 10 mm Measuring ranges DM 25 ON 50 Measuring span 120 7200 Fh 5900 25000 lh Start up 80 lh 500 Fh Medium aqueous liquids 1 96 FS re E 0 5 F full scale Optical display LC Display 2 x 16 characters d
74. ker med kabel og monteringsbeslag E3X DA11 N E3X DA21 N OMRON E3X DAB11 N E3X DA41 N E3X DA N 5 Teach er udf rt nar niveau displayet skifter fra r d til gr n Niveau displayet vil vise den aktuelle v rdi senere Gr n 6 S t mode switchen tilbage p RUN Cm RUN Kabel se note E3X DA51 N Bu To 2 4 mm dia 4 1 Mounting Holes Two M3 T Monteringsbeslag SUS304 rustfri st l Note P E3X DA11 N DA41 N DAB11 N anvendes 4 3 leder PVC kabel med et ledertv rsnit p 0 45 mm P E3X DA21 N DA51 N anvendes 4 4 leder PVC kabel med et ledertv rsnit p 0 2 mm
75. m To make the programming environment in LabVIEW less messy much of the code is grouped together to create different sub Vi s Figure 3 16 shows the content of the sub VI called Calibrate In this sub Vi the different measurements are calibrated and converted from mA and V signals to engineering units The hierarchy window figure 3 17 shows the relationship between the different subVIs in the program Experimental testing and verification 21 Hald up 1 pa Splits the signals and calibrate ix thems E CALIBRATE x Hold up 2 D i lu Ar PL KLS values Pressure 1 33 feed inlet y 0 2342 Pressure 2 upstream valve Hell 2326 10 04651 mase 5 mcm Flowmeter water y 3 739 8 14 35 mmm bg pr Figure 3 16 The content of the sub VI named FPE Read ez FID Filter Figure 3 17 Section of the hierarchy window Experimental testing and verification 22 3 4 Data analyzing and filtering Previous work on the Miniloop by B rdsen included the implementation of a simple PI controller to stabilize the process This controller used the downstream pressure P1 as the controlling variable This measurement however can be hard to obtain in practical problems like offshore installations One of the goals of this thesis was therefore to expand the work done by b rdsen 3 by implementing a cascade controlle
76. model storkaas was able to produce tuning parameters that would stabilize the given system Figure 3 38 show that the cascade controller with the tuning parameters given in table 3 8 stabilizes the system The set point for the pressure was 0 010 barg 0 15 P2 Setpoint 5 DI 1 5 8 005 Zt i j RK NN ai 0 UU 0 5 10 15 20 25 30 time min T 6 _ E aM 4 _ Wa 2 f 2 QUI 0 5 10 15 20 25 30 time min 1 SS 7 bifurcation point BS 2 5 _ gt gt 0 0 5 10 15 20 25 30 time min Figure 3 38 Cascade control with W in inner loop and P2 in outer loop model Experimental testing and verification 43 The system start in open loop with severe slugging and the controller is turned on after 70 s The controller forces the pressure up by closing the valve to an opening lying within the stable area Then the pressure is slowly driven towards the reference value as the system enters the unstable are After 30 minutes of simulation the pressure reaches its reference value As can be seen from the actuator usage the system 1s stabilized in the unstable area because the valve opening is at a value corresponding to slugging for open loop Table 3 8 Control parameters Inner loop Outer loop K 10 bar 0 02
77. n only water was pumped through the system a simple conservation consideration meant that the flow meter at the inlet would show the liquid flow through the valve Water was pumped through the system the actuator position was altered and the corresponding pressure drop over the choke valve and liquid flow into the system were recorded For pure liquid flow the density in equation 2 1 could be dropped since p 1 25 ite es 5 a KE D o o 1 5 Tp _ o o Q a 0 51 4 0 is 0 10 20 30 40 50 60 70 80 90 100 valve opening 0 2 ri 0 15 4 E D 0 1 4 2 D Zeee 0 05 4 0 0 10 20 30 40 50 60 70 80 90 100 Valve opening 9 Figure 3 23 Pressure and flow vs valve opening Experimental testing and verification 26 As can be seen from figure 3 23 valve openings above 30 seems to have little or no effect on the pressure or flow rate This suggested that the valve were oversized If equation 2 1 were to be valid the relationship between Q and zVAP had to linear However as can be seen from figure 3 24 this was not the case 4 Q l min 8 3 0 002 0 004 0 006 0 008 0 01 0 012 0 014 0 016 0 018 Z SQRT dP Figure 3 24 vs Z SQRT AP In an attempt to determine the valve characteristics two different approaches were tested The first was to try and fit the experimen
78. n the system is stable figure 3 21 contained large Tailor shaped bubbles The large gas volume in between the liquid bulks caused the data registered by the slug sensor to vary between liquid and air This in turn caused the flow estimate to oscillate Changing the piping to one with a bigger diameter may change the flow pattern during stable flow If this could be achieved the slug sensor would be able to estimate a more stable and continuous flow rate o The simplest solution would be to replace the slug sensors with a measuring device more suited for the task A capacitance meter has been implemented with great success by Kaasa 5 The capacitance meter gives an accurate value off the hold up and if used in series the delay can give the slug velocity e The cascade configuration with mass flow in the inner loop and the pressure drop over the choke valve failed to stabilize the process even though it should be possible according to the theory Perhaps a more advanced MISO controller would prevail where the cascade configuration failed In 6 an advanced multivariable H design were developed and tested and compared with a cascade configuration It concluded that the MISO H controller outperformed the cascade configuration The controller could handle large uncertainties up to 80 Perhaps this would allow it to stabilize the process even with the current measurement setup Conclusion 45 5 Conclusion By fitting the simplified model
79. nd variables can take 12 4 4 Writing the data to a file The miniloop program will automatically write the selected data to a txt file when the program is stopped using the large stop button The data will be stored in the following format Table 4 1 Format of stored data t msek S V g P D ar g W estimated Z l min kg min To store additional data to file use the following procedure 1 Open the block diagram and locate the write to file function in the right most section of the diagram All the data will be collected in the yellow block and collected in one array Its then send the the little white block and written to file upon termination of the program Figure 4 3 Create a free node 2 By clicking the yellow box and holding you can increase the length of it allowing more data stream to be added to it The box in figure 4 3 has been increased to allow 4 more data streams 3 The next stage is to locate the data you wish to store to the file Then simply wire it to an available node on the yellow box figure 4 4 The additional data stream will now be stored to the same file Figure 4 4 Wire the data 13 5 Maintenance The Miniloop do not require much maintenance However the user should be aware of a few things 5 1 Reservoir tank The user has to make sure the liquid level in the reservoir tank doesn t get to low From figure 5 1 one can see that there ar
80. obvious as it was simple The colouring matter used in the liquid was permanganate This gives the liquid a red colour However the property of a red substance 1s that it will absorb all light from the colour spectrum except red By taking the fact that the optical sensor used red light as its source into consideration the solution was obvious The water in the loop was changed and a blue colouring matter was added This gave the desired response on the sensors as can be seen from figure 3 19 Slug sensor STT TEPPE y Figure 3 19 The slug sensor after the colouring matter was changed By adding more colouring matter the amount of light absorbed by the liquid would increase and the lower value corresponding to pure liquid was reduced from 3 Volt to 2 volt The sensors digital display showed a value ranging from 0 4000 corresponding to 1 5 V where the value 1500 represented pure liquid The lower boundary for the sensor output was changed to 1700 making 1700 pure water correspond to 1 V and 4000 pure air to 5V In effect this cuts off all values below 1700 including most of the spikes figure 3 20 Slug sensor 5 0 Ss ME m Sg i H dad Figure 3 20 The slug sensor after signal scaling The slug sensor now gives a lot of information about the slug cycle From figure 3 20 a value of 5 indicates that there 1s only air in the pipe A slug 1s building up in the raiser and when the downstream pressure
81. onstration and control of multiphase production wells NTNU July 2003 6 Haukelids ter B advanced control Robust control of pipeline riser slugging NTNU june 2004 7 Skogestad S Postlethwaite I Multivariable Feedback Control Analysis and Design John Wiley amp Sons Ltd second ed 2001 8 Havre K Stornes K and Stray H Taming slug flow in pipelines ABB review 4 2000 pp 55 63 9 Matlab model simplified slug model aviable at http www nt ntnu no users espensto SlugModel 10 Seborg D E Edgar T F and Mmellichamp D A Process dynamics and control John Wiley amp Sons Ltd 1989 11 Sinnott K Coulsons amp Richardson e Chemical Engineering Volume 6 Butterworth Heinemann Third edition 1999 12 National instruments http www ni com aap Appendix A 47 Appendix A A 1 Miniloop Experimental data The experimental data used to create the bifurcation diagram for the Miniloop were stored and are shown in figure A 1 A 13 The figures show the upstream pressure and downstream pressure P2 vs time To create a cleaner visualization of the pressures a Savitzky Golay filter was added A 2 Savitzky Golay filters The data used to create figures to A 13 are stored as the miniloop program received them i e without additional filtering The filter applied to the data in had a frame size f 21 and a polynomial order k 3 savit
82. or d i Ion Figure 3 13 The front 1 for the labview program miniloop The front panel has three main areas of interest First you have the charts used to visualize the measurements like pressure drop valve opening flow and hold up The top chart displays the downstream pressure while the second one displays the upstream pressure If anti slug control is applied the mentioned charts will display the relevant set point The third chart from the top plots the flow of water into the system and an estimate of the flow through the control valve If a cascade controller 15 applied it will also show the relevant set point The slug sensor results are plotted at the bottom This measurement plots the filtered signal received from the optical sensors The PID control is located at the lower left corner of the screen This is were the user chooses witch control structure to use The loop is set to no control by default but by clicking it you can choose the following control structures from the pull down menu No control PI control with P1 as the controlled variable Cascade control with mass flow W as the main variable and P1 in outer loop Cascade control with mass flow W as the main variable and P2 in outer loop Experimental testing and verification 18 The tuning parameters for the different controllers are also located here witch means the user can change them by simply entering the new value In the upper left corn
83. otalizer T41 Unions imperial 33 0 4 20 pulse output Batch controller BRZ 2x Relay outputs Body material Ref no Supply volage PVC LI 1 24 V DC 20 Order exampe a ERR ER EREECHEN Type Nominal size ref no Body configuration ref Connection ref na Body matenal ref no Seal material ref no Display position ref na Functional profile ref na Supply voltage Display position with regard to flow direction E E La 2 Connection X1 4 GND for totalizer and 2 24 batch controller A Connection X2 1 GND Tor totalizer 2 Curent output 0 4 20 m l 3 Impulse output PRP 4 NC Connection X2 1 Common 1 for batch controler 2 Relay output BOty 2 3 Relay output BFast A c ETO VALVES ACTUATORS AND CONTROL SYSTEMS Flow Teknikk as Mass Flow Meters and Controllers for Gases Mass STREAM EG EE M W Instruments Your partner Key Facts M W Instruments was founded in 1988 and has always specialised in mass flow meters and controlers In 1995 M W Instruments was the first company to introduce the direct measuring principle for thermal mass flow meters with the sensor following the constant temperature anemometer principle In 1997 M W Instruments joined the Bronkhorst Group zs Content Key Facts Working Principle sss seite 2 _ 5 5
84. ow pressure drop Unique all stainless steel sensor Pressure and temperature indepen dent metering system The 62xx series working principle is shown on page 2 Main advantages of these instruments are Installable in virtually every position No inlet pipes needed 62xx series Optional with integrated flow display Optional with totalisation No maintance needed Two body materials on stock others on request For lower flow values the by pass measu rement principle is applied Model D 6210 MFM Be Es ee 1 S 8r 5 LES ED z Standard Flow Capacities Mass Flow Meters Flow capacitiy Air M odel intermediate ranges available D 5110 12 A A A 0 005 0 1 l min Air 22 0 010 0 2 l min Air ST ee a p 9 ee re D 6210 AA 53 0 25 50 Il min Air Fr eee 35 D 6230 AA 24 10 200 erte 29300 min Air Display optional 15 5 0 100 0 Air re 1 D 6250 15 5 0 100 0 Il min A
85. played in its entirety but figure 3 14 shows a simplified sketch of the dataflow inside the block diagram A while loop encompass most of the program code The field point modules will provide the while loop with the data sampled from the measurement devices Inside the loop the data 1s calibrated filtered and modified to provide the measurement needed for control purposes The box labelled control structure is a case structure that contains the different control structures mentioned in chapter 3 3 1 The code inside the while loop will provide the control structure with the measurements needed and the actuator position will be send back to the field point modules through the while loop If no control is active the control structure will return the actuator position set by the sliding bar in the front panel While loop Control structure Figure 3 14 The data flow inside the block diagram Figure 3 15 shows about 1 4 of the block diagram associated with the miniloop program To view the diagram in its entirety the reader should open the miniloop program on a computer with Labview installed and access the block diagram ctrl e The program can be found on the cd that accompanied this thesis The partial frame seen in the lower right corner is the control structure in figure 3 14 Experimental testing and verification 20 Weien alve position posit poet Valve position DELE Figure 3 15 Section of the block diagra
86. r Previous work by Storkaas 4 suggested that a cascade controller using mass flow W or volume flow Q as the controlled variable could stabilize the process A measurement of the mass or volume flow would have to be estimated from other available measurements like the slug sensor Before such an estimate could be obtained the slug sensor had to be analyzed further 3 4 1 Slug sensors The slug sensors had been installed by Bardsen when the loop was build but no work had been done on analyzing the signals they produced The original signals figure 3 18 consisted mainly of a constant signal interrupted with spikes The purpose of the slug sensors was to measure the hold up and it soon became apparent that the signals had to be treated further if any useful information were to be gained from them JL M EE EE Figure 3 18 Initial readings from the slug sensors As mentioned earlier the slug sensors are based on fibre optical technology Two separate optical cables are connected to a sensor Light 1s emitted from the sensor and travels through one of the cables As the light exits the first cable it will travel through the medium to be measured The second cable is mounted on the other side of the medium directly opposite the first cable The light emitted from the first cable travels through the medium and is returned to the sensor through the second cable The sensor will produce a signal between I and 5 volts depending on the amount of li
87. r temperaturen m lt p fiberspidsen Tilladelig b jnings radius E32 L56E1 E32 L56E2 OG b E32 L24L 10 mm E32 L25L E32 L25 25 mm 7 C E32 L25A E32 R21 E39 R3 tilbehor Z 5 E32 R16 E39 R1 o tilbehor OMRON E3X DA N E3X DA N Specifikationer Tekniske data Fiberforst rker Standardmodel Med analog udgang M rkeaftaster Med 8 1 8 PNP udgang EDAM N ESXDASN ESKDAMV _ 12 til 24 VDC 10 ripple 10 max Str mforbrug Normalt 960 mW max str mforbrug 40 mA max ved 24 VDC forsyning Eco Mode 720 mW max str mforbrug 30 mA max ved 24 VDC forsyning Display sl et fra 600 mW max str mforbrug 25 mA max ved 24 VDC forsyning Kontrol udgang ON OFF udgang NPN PNP afh ngig af model ben collector belastningsstr m NPN aben NPN PNP afh ngig 50 mA max Restsp nding 1 V max Light ON Dark ON kan collector af model aben veelges belastningsstr m collector 50 mA max belastningsstr m restsp nding 1V 50 mA max Light Restsp nding 1 V ON Dark ON kan max Light ON Dark v lges ON kan v lges min Beskyttelse af kredsl b Omvendt polaritet kortslutning af udgang gensidig p virkning Responstid Super high speed mode 0 25 ms for drift eller reset Standard mode 1 ms for drift eller reset Super long distance mode 4 ms for drift eller reset F lsomhedsindstilling Teach eller man
88. r use is acceptable It s constantly making small adjustments to keep the system stable and at the stationary unstable solution It 1s evident that the system 15 stabilized in the unstable region since the valve 15 operating at a valve opening that lies in the unstable are for open loop This can be seen from the bifurcation diagram figure 3 29 As excepted the slugging reappears quickly after the controller is turned off In light of the control objective the pressure controller 15 performing very well 0 16 d d d Pressure P1 f d 0441 e Setpoint a HUE BN 0 121 918 0 08 P1 barg 0 06 0 time min o o Oo CO T Valve postition 0 2 hed 0 M 0 1 2 3 4 5 6 time min Figure 3 33 Performance of a pressure controller on the miniloop Experimental testing and verification 36 simulation was also run on the simplified slug model figure 3 34 using the same parameters for the gain and integral action as above The set point for the controller was also the same 0 115 barg As above the pressure 1s represented with the blue line and the red line indicates the set point The simulation starts in open loop resulting in severe slugging After 2 minutes the controller 15 turned on After an additional 3 minutes the controller 15
89. ra Dupont Company Applikation Tasteafstand mm 1 Udseende Forst rker p Reflektor E39 R3 Ei Reflektor E39 R1 Egenskaber E3X DAI Taster i et stort omrade ud fra siden Aftastning af areal Aftaster med refleksbrik E3X DAL Aftastning af transparente emner 10 to 250 10 to 250 10 to 250 M6 E3X DAI Aftastning af transparente em ner IP66 Fast taste afstand E3X DAL Positionering af glas Aftastning af sma h jdeforskelle Omgivelsestem peratur imellem 40 C til 105 C IP50 E3X DAL nd Kontrol af Montering p r r 8 E3X DAL N Veeske E32 L25T 10 mm veesker E3X DAL Fast taste afstand Aftastning af sma h jdeforskelle IP50 E3X DAL E3X DAL Note 1 V rdierne for det mindste aftastbare objekt er opn et ved en ikke oplyst afstand 2 Det kan v re n dvendigt at udf re et eller to punkts teach for at samme resultat 1 Tasteafstand indikerer v rdier opn et med hvidt papir Standard objekt min tasteemne 5 Guld tr d 200 x 200 0 01 mm dia 200 x 200 0 01 mm dia Indikerer v rdier for standard mode Standard objekt min tasteemne 2 Guld tr d 300 x 300 0 01 mm dia 100 x 100 Indikerer v rdier for standard mode E3X DA N Tilladelig b jnings radius E32 D51 Indikere
90. realistic to achieve a good fit for both the slug regime and the stationary regime for the entire range of valve openings so some priorities had to be made during the tuning of the model If the controller 1s to work properly the actuator has to have a significant impact on the process For the high range valve openings the pressure drop over the choke 15 too small for the effect of a small change in valve opening to be significant This means that the only area of interest 1s that from medium to small valve openings This can be seen from the bifurcation diagram because it shows little to no variations from medium to high valve openings Secondly the unstable stationary regime is more important then the stable oscillatory regime The reason for this is that the goal is to achieve stationary flow with active control in the unstable area By avoiding the slug regime and operating at the stationary regime the focus 1s on where you want the process to be and not where you don t want it to be A good example of this was presented by Storkaas 1 2 If you are teaching someone to ride a bike you are teaching them how the bike behaves when they have mastered the balancing act the desired unstable operating point not how it behaves when it lies on the ground the undesired slug flow When taking these priorities into consideration the simplified model shows a good fit to the experimental data The model fit to the experimental data for the
91. rements In the previous chapter it was shown that a simple PI controller could stabilize the system using one upstream measurement P1 As mentioned earlier this measurement is in many cases not even installed In other cases it will often prove unreliable or unusable for control The controllability analysis in chapter 3 7 provided the basis for exploring other possibilities P2 as measurement The controllability analysis could not dismiss the pressure drop as possible measurement For this reason an attempt was made to stabilize the Miniloop by using a pressure controller with P2 as the measurement All attempts to stabilize the system proved unsuccessful and P2 was dismissed as a possible measurement for a stabilizing controller Since the controllability analysis in 4 had excluded P2 as a measurement it will not be treated any further here 3 8 3 Cascade control According to the controllability analysis both the total mass flow W and total volume flow Q were better suited for a stabilizing controller However stabilizing the system with one of these measurements would lead to an almost integrating closed loop system but a cascade configuration with a pressure measurement in the outer loop could solve this problem Because of the problems with obtaining an estimate of the volume flow rate chapter 3 4 2 the mass flow W was chosen as the measurement Two different cascade configurations were tested where the downstream and upstream pres
92. res are located inside within the case structure New control configurations can be included in the program by adding a new case and filling in the relevant code Below is an example of case 3 This is a cascade configuration that uses mass flow W in the inner loop and upstream pressure in the outer loop The measurements and set point is passed from the while loop to the case The controllers then calculate the valve position and sends it back to the while loop Fi The PID controllers The tuning parameters Set point and variable gure 4 2 Case structure the controller e ce olo ofofo eleau S KE wi 1 4 ke b i 4 ke b ke i ke 4 i 4 ke ke ke b i b ke b ke b ke 4 ke 4 i 4 ke b ke ke b ke b 4 ke b ke 1 ke b ke b i 4 ke 4 ke de ke b ke The actual PID controllers are located within the blue circles The parameters associated with the PID controller are the min max output signals the controller can take In case of the outer loop these parameters represent the values that the set point in the inner loop may take These are located within the red circles They actual values are set in the front panel The set points and measurements have to be converted from engineering units to percent of operating area this is done in the green circles The parameters set here correspond to the min and max values that the set point a
93. returned to manual mode The controller quickly stabilizes the system and the tracking performance Is very good The actuator use 15 minimal as it moves towards the valve opening corresponding to the set point of 0 115 barg It seems to stay stable at this value however if one had zoomed in on the graph one would have seen that the actuator 1s continuously making small corrections to hold the system stable 0 16 Pressure F1 0 14 Setpoint zx 0 12 m a 01 c3 e time min valve position 0 1 2 3 4 5 H D g 10 time min Figure 3 34 Performance of a pressure controller on the simplified model By comparing the simulation to the experimental data figure 3 33 it is obvious that the simplified model is in accordance with the experimental data The simplified model simulates the results obtained experimentally very accurately More importantly the controller in the simulation reproduced the stability results obtained experimentally from the lab scale Miniloop The biggest difference between the model and the Miniloop is the time it takes for severe slugging to reappear after the controller is turned off While this takes less then a minute for the Miniloop the model requires almost three minutes The reason for this is that the simplified model contains little noise causing the pressure to stay close to the reference value for a longer time Experimental testing and verification 37 3 8 2 Control with downstream measu
94. selector back to run mode Factory setting R Ldjes E In units of 100 from 0 Altemates 0 901 DU Q Altemates SS TEA at TEAGH Q Figure 5 5 Setting the lower limit for monitoring 16 5 3 2 Troubleshooting the slug sensor This chapter should solve the most common problems with the slug sensor Problem The slug sensor chart in the user interface 15 not taking values between I and 5 The slug sensor chart is showing values below 5 for pure air when it should show 5 Solution When there is only air in the tube no light should be absorbed and the sensor should show its max value Check the digital display on the sensor The display should show 4000 when only air is present If it shows less something is interfering with the light beam Make sure the sensor is not placed on a stained part of the pipe as stains may block some of the light moving it to a different location or changing the stained section of pipe will most likely solve the problem The cables may also be bent resulting in a failure within the cable Make sure the cable is running loosely and smoothly from the sensor to the bracket without to large angels The slug sensor chart is showing values higher then I for pure water when it should show 1 Solution Most likely the lower limit for monitoring is set to high or there 15 too little colouring matter in the water Check step 2 in chapter 5 3 1 5 4 Colouring matter
95. ss conditions The conversation factor introduces an additional hydrogene it is always recommended to make use of the optional gas error in abolute accuracy in order of calibration CF gt 1 2xCF in FS CF lt 1 2 CF in FS zum Fow profile sensivity Normally mass flow measurement principles are sensitive to varia The Mass STREAM Flow meters has been designed in such a way tions in the shape of the flow profile that there is always a fully developed flow profile in the metering section ans is thus virtually insensitive to changes upstream piping conditions mum Pressure drop The pressure drop of the instruments serie D 62xx is almost com parable to a straight run of pipe of the diameter ans 15 thus negible However to make the instruments insensitive to upstream piping configurations a number of mesh screens are required to condition 100 0 10 15 20 25 40 50 60 70 80 90 ltr ltr ltr Itr ltr ltr ltr Itr ltr ltr ltr ltr ltr ltr AP mbar 100 200 300 flow air l min In some applications a very low pressure drop 15 required To fulfil these inquires we are able to build a special version of our instru ments where the pressure drop 15 extremely low H 5 the flow profile These meshes create certain pressure drop as can be seen from the table set out below For further informations please contact factory
96. sure were used for feedback purposes Experimental testing and verification 38 3 8 4 Mass flow W and upstream pressure P1 Figure 3 35 shows that a cascade controller using mass flow W in the inner loop and upstream pressure P1 in the outer loop will stabilize the system The system 15 started m open loop with severe slugging and the controller 1s turned on after 2 minutes After an additional 3 2 minutes the controller 15 switched back to manual The inner loop uses a pure proportional controller while the outer loop uses both proportional and integral action The parameters for the gain and integral action are listed in table 3 6 Table 3 6 Tuning parameters Inner loop Outer loop K 0 8 bar 15 min kg 30 5 The two upper charts show the pressure vs time and the mass flow vs time The red line indicates when the controller is active and represents the set point of 0 115 barg for the outer loop The set point for the inner loop 15 provided by the outer loop The actuator use 15 displayed in the bottom chart 0 16 rf P1 e TIT 2 0 12 E Q M 0 08 0 06 0 1 2 3 4 5 6 7 8 9 10 time min 6 E HELAN hid ha Lil nA j p wn UI UL D 2 d AE LL 0 LU U
97. t in cracks P mE VE H d M E A gt 4 D oh i Figure 5 2 Buffer tank 14 5 3 The slug sensor There are quite a few things that can cause the slug sensor to fail The slug sensor should always be checked to make sure it is operating as intended before an experiment is started The easiest way to check the sensor is to make sure the slug sensor chart in the user interface miniloop program front panel figure 3 1 15 taking values between 1 and 5 It should be I when measuring pure water and 5 when measuring only air If this is not the case something is causing the sensor to malfunction Take note that the problem may not be with the sensor itself but with the Miniloop This chapter will show the user how to reconfigure the sensor from start However if the sensor is malfunctioning consult the trouble shooting in chapter 5 3 2 Figure 5 3 shows the actual sensor and the different settings on it Setting Buttons TEACH MODE Lock Button Lavel Display Je una M il Operation Indicator orange Operation Mode Selector ON when output is OWN Use to switch between OFF when outputis OFF Mode Selector Light ON and Dark ON modes Use to select SET ADJ or RUN mode Figure 5 3 The sensor settings 5 3 1 Calibrating the slug sensor Step I Reset the sensor to default settings as shown in figure 5 4 Set the mode selector to SET E
98. tal data to equation 3 2 by a linear parameter estimation method a and the second was to estimate f z from equation 3 3 by plotting OU AP against z method b Both methods are described in more detail in appendix A A Q k z JAP 3 2 f zNAP 3 3 Both methods proved unsuccessful when trying to fit it to the entire data range In order to get an estimate of the flow a purely mathematical approach had to be abandoned For control purposes the priority had to be to obtain a satisfactory flow estimate in the area with small valve openings As mentioned earlier valve openings above 30 seemed to have little or no effect on the flow rate The two approaches mentioned above were therefore attempted on the data set below 30 valve opening The best results were obtained by method b when a fifth order polynomial was used to describe f z Experimental testing and verification 27 e Datapoints Poly Datapoints 0 1 0 15 0 2 0 25 0 3 Valve position 2 Figure 3 25 Fitting the f z to the datapoints The fifth order polynomial fitted to the data using the least squares method 15 f z 702232 613502 1919173 2705 47 229 44 3 4 Combined with equation 3 3 this estimated the flow through the valve for valve openings below 30 To make the equation valid for valve positions above 30 a linear relationship were assumed between the flow rate and the pressure drop by setting f z f 0 3 for all valve op
99. te at a higher valve opening then what was possible for the PI controller before the system would go unstable By lowering the set point even more to 0 1 barg the process would operate at valve openings over 30 This would allow even better ol recovery in real life applications in the oil industry Experimental testing and verification 40 Figure 3 36 show the same case simulated using the simplified slug model The parameters are the same as those given in table 3 6 The simulation starts in open loop and the controller is switched on after 2 minutes After an additional 5 minutes the controller is returned to manual The controller in the simulation also reproduced the stability results shown experimentally above The pressure 1s effectively stabilized even though the model takes a bit longer to reach the reference value Like the previous case it takes more time before the slugging reappears in the model compared to the Miniloop All in all the simulated response 15 in accordance with the experimental data F Setpoint time min Massflow c Bem MR 0 1 2 3 4 5 B 9 time min Valve position c DN 0 1 2 3 4 5 H D g 10 time min Figure 3 36 Cascade control with W in inner loop and P1 in outer loop Model Experimental testing and verification 41 3 8 5 Mass flow W and downstream pressure P2 In light of the results obtained in the previous chapter the prospect of achieving satisfactory
100. tforms The trend towards more satellite wells means that the multiphase flow has to be transported over greater distances In addition to the increased length greater depths provide additional challenges for multiphase transport and control The pipeline riser systems needed to transport the oil to the production facilities gives rise to an undesired flow regime known as slug flow Slug flow usually occurs when you have a low point in the pipeline topography The liquid will accumulate at the low point blocking the pipe and result in the forming of a liquid slug The slug will continue to grow until enough upstream pressure has developed to overcome the weight of the liquid slug These slugs can grow very large and often cause severe problems when they reach the production facility Severe slugging can in the worst case lead to a plant shutdown More frequently the large and rapid variation in flow leads to poor separation and unwanted flaring Severe slugging can be avoided by increasing the pressure drop over the top side choke valve Early solutions involved closing the top side choke valve to avoid the slugging This solution is far from optimal as it will result in a reduced oil recovery Other solutions include installation of slug catchers By applying active feedback control it 1s possible to stabilize the flow at a pressure drop that would normally lead to severe slugging This reduces the need for additional topside equipment and allows a hig
101. thout the complexity of traditional development tools 12 The user interface in LabVIEW is called a Virtual Instrument VI These VI s have to be made for the specific measuring set up in this case for the Mini Loop The programming language used in LabVIEW 1s called G and 1s a graphical drag and drop programming This programming language is based on C and LabVIEW supports additional code both in C Visual Basic and Matlab Scripts LabView consists of three main parts The front panel is the interactive user interface of a VI so named because it simulates the panel of a physical instrument The front panel can contain knobs push buttons graphs and many other controls user inputs and indicators program outputs Data or control can be input by a mouse or a keyboard and results can be viewed by the program on the screen A LabVIEW includes a wide array of visualization features including tools for charting and graphing This makes it easy to visualize data The user simple drags and drops the desired controls or indicators from the build in control palette to the front panel Figure 3 12 Labview Front Panel The second part of LabVIEW is the block diagram Its in the block diagram that the data is acquired analyzed and generated While the front panel is the user interface the block diagram is the VI s source code It is constructed in LabVIEW s graphical programming language G The block diagram is the actual executable
102. ting valves V1 and V2 The pipeline system is constructed of several connecting sections of transparent plastic tubes The pipeline is meant to imitate the pipeline topography where gravity induced slugging occurs which is a low point connected to an inclined section of the pipe At the top of the riser the multiphase flow passes the control valve before it enters the separator At this point the air is released out of the system through an open hole in the separator while the water is returned to the reservoir To monitor the behaviour of the system a combination of pressure measurements and slug sensors are used The measured signals are transmitted to a computer through the FieldPoint FP modules where they can be analyzed stored and manipulated Table 2 1 List of equipment Denote Equipment FT W Rate meter for water Gemu 3021 FT A Rate meter for Air PI Pressure sensor MPX5100DP Feed inlet 2 Pressure sensor MPX5100DP Valve P3 Pressure sensor MPX5100DP Separator air outlet SI Slug sensor E3X DA N 52 Slug sensor E3X DA N PU Pump Eheim 1060 WT Reservoir BT Buffer tank ST Separator CV Control valve Consult table A 1 for more info about the distributors and prises More detailed information about the different equipment can be found in appendix B The different equipment will be briefly discussed below The rate meter for water figure 2 1 is placed in front of the mixing point of water and air The digital display shows t
103. to the experimental data gathered from the Miniloop it has been shown that the model is a good representation of the process This can be seen from the good fit of the bifurcation diagram The controllability analysis shows that the best measurement alternative for a stabilizing controller is the upstream pressure P1 Both the downstream pressure P2 and pm are badly suited as a stabilizing controller The controllability analysis show that the total mass flow or total volume flow are better alternatives Stabilizing the system with one of these two measurements will result in an almost integrating process But a cascade configuration with flow in the inner loop and pressure in the outer loop could solve this problem A simple PI controller using the upstream pressure P1 as measurement stabilized the system Two different cascade configurations were also tested One used the upstream pressure P in the outer loop while the other one used the downstream pressure P in the outer loop Both cascade configurations used the mass flow W in the inner loop The cascade configuration with P in the outer loop managed to stabilize the process around a valve opening that would result in severe slugging during open loop The low gain in the inner loop means that the actual stabilizing task is done by the outer loop The inner loop merely acts as a filter The cascade configuration with P in the outer loop also stabilized the system but around a valve opening
104. tur imellem 40 C til 200 C Modst r 150 C 2 Omgivelsestemperatur imellem 40 C til 150 C Vinkelaftastning modst ar 150 C 2 detektion af sma emner Omgivelsestemperatur imellem 40 C til 150 C Ideel til aftastning af etiketter o lign Aftastning af lille h jdevariation Vinkelaftastning aftastning af lille h jdevariation 4 f lerhoveder for genkendelse af former og h jder Taster i et omr de p 30mm Taster i et omr de p 10 mm lang tasteafstand Taster sm emner i et omr de p 11 mm IP50 OMRON Forst rker Tasteafstand mm Standard V rdier i parantes ved objekt 4 anvendelse af E39 F1 linse mindste uigennem sigtige emne E3X DAL ON E3X DAL ON E3X DAL ON E3X DAL ON E3X DAL ON E3X DAL N Ru E3X DAB11 N E3X DAL N E3X DAL ON E3X DAL ON E3X DAL N E3X DAL ON E3X DAL ON E32 T12F E32 T14F E32 T61R E32 G14 E32 T22S E32 T24S Q E32 T16W W E32 T16 E32 T16P St rrelsen p standardobjektet er den samme som fiberens diameter anvendes linse er det linsediameteren E3X DA N Tilladelig b jnings radius 40 mm 10 mm
105. ual metode Forsinket frafaldstimer 0 til 200 ms 0 til 20ms justeres i spring af 1 ms og 20 til 200 ms i spring af 5 ms Automatisk Sikrer ensartet lysmaengde i hele levetiden Sikrer ensartet str m kontrol lysmaengde i hele levetiden Displayv rdierne kan resettes til 0 hvis nsket negative v rdier kan ogs vises Mulighed for at stille tilbage til fabriksindstilling Skalering af vre og nedre gr nser kan analog udgang skaleres i enheder af 100 Display Udgangsindikator orange 7 segment display r d og gr n V lg mellem 3 visninger 12 bit visning procent visning eller analog bargraf visning Andre display visninger V lg mellem Normal Peak h j Peak lav visning Display orientering V lg mellem normal og omvendt visning Justering af optiske akser Mulighed for at indstille optiske akser ved hj lp af hyper flashing funktion Omgivende belysning Gl delampe 10 000 Ix max Sollys 20 000 Ix max Omgivelsestemperatur I gruppe med 1 til 3 forst rkere 25 C til 55 C I gruppe med 4 til 11 forst rkere 25 C til 50 C I gruppe med 12 til 16 forst rkere 25 C til 45 C uden tilisning eller kondens Opbevaring 30 C til 70 C uden tilisning eller kondens Drift og opbevaring 35 to 85 uden tilisning eller kondens 20 MQ min ved 500 VDC 1 000 VAC ved 50 60 Hz i 1 minut 10 til 55Hz 1 5 mm dobbelt amplitude eller 300 m s2 ca 30G i 2 timer i retning X Y og Z 500 m s
106. udtzen AS 1550 the control valve DC DC Select P O BOX 160 N 1378 Nesbru Phone 47 66 98 33 50 3 x Pressure sensors MPX5100DP Silica Avnet Nortec AS 796 P O BOX 63 N 1371 Asker Phone 47 66 77 36 00 Pumpe Eheim 1060 Petshop at city syd 1566 3x Optic sensors E3X DA N Omron 2825 P O BOX 109 Bryn N 0611 OSLO Phone 47 22 65 75 00 3x Tanks 9000 Tubes 4000 19 Appendix B Equipment manuals The equipment manuals are large and comprehensive Because of this not all of them will be appended to this user manual However the table below will list where they can be found The manuals for the equipment listed in table B 1 can be found in appendix B in 1 while the manuals listed in table B 2 can be found in this appendix Table B 1 Reference to equipment manuals Equipment Pressure sensor Signal transducer control valve FieldPoint A1 Module FieldPoint AO Module FieldPoint Terminal base Table B 2 Reference to equipment manuals Equipment Rate meter for liquid Rate meter for gas Optic sensor In Danish Appendix Bl B2 B3 20 L3 EML Flow transmitter Construction An intelligent flow transmitter which can ba used for measuring liquid inert and corrosive aqueous media The keypad at tha front of the unit enables simple setting of measurement units required display values etc Features gt High resolution turbina flow measurement Medium wetted parts in plastic and ceramics Gan be used as total
107. ure 3 2 is placed in front of the mixing point of water and air The digital display shows the rate of water in l min It provides a signal between 4 20mA depending on the rate of flow which 15 send to the computer Figure 3 2 Rate meter for water The rate meter for air Figure 3 3 is placed in front of the mixing point of water and air It has a digital display that shows the rate of air in percent of its operating area witch 15 0 2 2 l min The rate meter also provides a signal between 0 5 V which is send to the computer Figure 3 3 Rate meter for air Experimental testing and verification 12 The pressure sensors Figure 3 4 is one of Motorola s differential pressure sensors that delivers a signal between 0 2 4 5 V The relationship between voltage and pressure 1s linear and its operating area is between 0 100kPa Figure 3 4 Pressure sensor The slug sensors Figure 3 5 are fibre optical sensors Each slug sensor is made up of two fibre optical cables connected to a sensor The light emitted from the senor will travel out through one of the cables and back through the other The device will provide a signal between 1 5 V depending on how much light is transmitted between the two cables Figure 3 5 Slug sensor The pump used is a standard aquarium pump It can deliver up to 38 l min and work against a head of 3 1 m Special care must be taken to make sure it doesn t pump air as this can damage t
108. uring these variations as the reference The air flow measurement is dependent on the temperature inside the measuring device This means that the measured flow rate of air will drop during the first minutes after start up as the temperature inside the device stabilize it self Because of this its not recommended to initiate any experiments until the measurement is stable approx 5 10 mins 3 2 User interface The Miniloop is controlled by a computer through the user interface teg 00722 0 005722 1 000 3 000 10110 10 100 1 20 1 15 0000 d foo 1919 00 28 125 0 400 10 020 gt 1 ge Joo0 120 d s 100 iis 10 0 Figure 3 1 User interface for the Miniloop The front panel has three main areas of interest First of you have the charts used to visualize the measurements like pressure drop valve opening flow and hold up The top chart displays the downstream pressure while the second one displays the upstream pressure If anti slug control is applied the mentioned charts will display the relevant set point The third chart from the top plots the flow of water into the system and an estimate of the flow through the control valve If a cascade controller is applied it will also show the relevant set point The slug sensor results are plotted at the bottom This measurement plots the filtered signal rece
109. very 3 6 The simplified slug model The model used to describe the riser slugging behaviour 1s not a partial differential equation system but a simplified bulk model The model has only three states the mass of gas behind the slug the mass of liquid in the slug and the mass of the gas in front of the slug Riser slugging as described by the simplified model can be seen in figure 3 28 Experimental testing and verification 30 mG2 P pc GLT h gt H SW 5 0 l LII W in V wW al Mg Vor War Lin st Hj not Figure 3 28 Model characteristics with important parameters The model was tuned to fit the experimental data as shown in figure 3 29 The red lines 1s the data from the simplified slug model while the blue dotted lines shows the experimental data The red dashed line indicates the presence of an unstable stationary solution at the same choke valve openings as those corresponding to severe slugging 0 25 l Storkaas simplified model experimental data 0 2 V P1 barg 0 1 _ _ l L 0 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 1 valve opening z P2 barg 0 02 0 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 1 Valve opening z Figure 3 29 Bifurcation diagram for the simplfied slug modell Experimental testing and verification 31 It is un
110. viour Stationary pressure balance between riser and feed section Simplified choke model for gas and liquid leaving the riser Constant system temperature Theory 9 2 4 Controllability analysis The following theory is found in 7 2 4 1 Limitations imposed by RHP zeros and unstable poles A linear dynamic system can be represented as x Ax Bu The system 1s stable if and only if all the poles are in the left half plane LHP Rel 4 0 Vi Unstable poles can be stabilized by feed back control Right half plane RHP poles impose a lower bound on the bandwidth w for a system 7 provides the following lower bound for RHP poles W 2p 2 3 and for imaginary poles W 1 15 p 2 4 RHP zeros results in an inverse response and impose an upper bound on the bandwidth for any system using feedback control When a system is using feedback control the closed loop poles will approach the open loop poles as the gain approches infinity This makes the system unstable and limits the bandwidth for high gains The bound on the upper bandwidth for a system with real RHP zeros Is W SW 2 5 And for complex zeros lz 4 Re z gt gt Im z W xW 12 2 8 Re z Im z 2 6 Z Re z lt lt Im z RHP zeros close to the origin of coordinates impose the largest constraints on the bandwidth Control is more difficult if the zeros are located close to the origin of coordinates compared to zeros close to the imaginary
111. w through choke ennen nen ereerrreeeerereee 8 23 Mode 8 2 4 NNN 9 2 4 1 Limitations imposed by RHP zeros and unstable poles 9 3 Experimental testing and verification u ddssssseeeeeeeeeerereeeenenrerereernnsrerereee 10 3 1 LT 10 3 1 1 FP vvs 11 3 1 2 Changes tothe EE 14 32 Dam Flow ANG Dala NN 15 32 1 SE NN 15 3 3 EE EE NN X 16 3 3 MK 1O0P FOM PL NN 17 3 3 2 ke een Ree Tee 19 2d De NNN NG 22 3 4 1 SNS OLS EE 22 3 4 2 Estimating the flow through the choke valve 25 3 5 sv 29 3 6 ThemibDiuted Sip model vvs iE EE E v EIE ea DO Rs 29 3 7 Clan alver ta 33 Dee PR ER 35 3 8 1 Control with upstream measurements rrrrrrrrnrrnnnnnnnrrrrrrrnrrnnnnnnenrrrrnnnnnnnnnneeeee 35 3 8 2 Control with downstream measurements rrrnnnnrrrrrnnnnrrrrrrnnnnnnnnnnnnnnnrrrrrnnnnnnnne 37 3 8 3 accessed 37 3 8 4 Mass flow W and upstream pressure Pl 38 3 8 5 Mass flow W and downstream pressure PI 4 3 9 User manual MIN OOD iud oae oe er 45 4 Xuture WOEk ao epe NR 44 gt MONET ES E 45 DEN den 47 Appendix B User Manilas 60 Table of figures Figure 2 1 Figure 3 1 Figure 3 2 Figure 3 3 Figure 3 4 Figure 3 5 Figure 3 6 Figure 3 7 Figure
112. ystem in the simulations The reason the cascade controller failed to stabilize the Miniloop was because of the disturbances and the noise picture associated with the flow measurement The result obtained from the experiments verifies the simplified slug model as a useful tool for control purposes Acknowledgement Various people have been of assistance during the work and experiments on the Miniloop as well as in the writing of this final report I would therefore wish to thank the following people for their invaluable help and support e Supervisors Espen Storkaas and Heidi Sivertsen for being there and for all the invaluable help they provided during the experimental phase of the thesis A special thanks to Espen for providing and assisting me with the simplified slug model Professor Sigurd Skogestad for all the tips and hints provided Ingvald Bardsen for help in understanding the Miniloop Torgrim Aas for demonstrating the Miniloop build at Statoil Ole Ivar Hovin for reparing equipment that broke Jan Ole Sundli for updating the field point modules Table of contents MEN DD PM 4 1 1 BOL 4 2 SU roa erm 4 1 3 SCOPE OF LMS IIE 5 2 Then svaner 6 2 1 MR 6 2 Gravity induced SIUC O11 ote nr uae mates Ee 6 22 e 7 221 Single phase liquid flow through chokes I 222 Gas Flow tee 7 22 9 Multiphase flo
113. zky Golay filters are optimal in the sense that they minimize the least squares error in fitting a polynomial to frames of noisy data y sgolayfilt x k f applies a savitzky Golay FIR smoothing filter to the data in vector x If x is a matrix sgolayfilt operates on each column The polynomial order k must be less then the frame size f witch must be odd If k f 1 the filter produces no smoothing 100 96 Valve opening 0 16 0 14 S 012 a oi 0 08 0 06 0 20 40 60 80 100 120 time s 0 15 D S 0 1 CN 0 05 X MAL uL ux 0 20 40 60 80 100 120 time s Figure A 1 Open loop data for z 1 Appendix A 48 80 Valve opening 0 16 gt 0 14 2 0 12 a 0 1 0 08 0 06 0 20 40 60 80 100 120 time s 0 15 o 3 0 1 N 0 05 0 0 20 40 60 80 100 120 time 5 Figure A 2 Open loop data for z 0 8 60 Valve opening 0 16 3 0 14 S 0 12 a 0 1 0 08 0 06 0 20 40 60 80 100 120 time s 0 15 D m 0 1 N a 0 05 NVS 0 20 40 60 80 100 120 time s Figure A 3 Open loop data for z 0 6 Appendix A 49 40 Valve opening 0 16 0 14 0 12 0 1 0 08 ud 20 40 60 80 100 120 P1 barg time s 0 15 S 0 1 y 0 05 EA LA ALAN AN 0 20 40 60 80 100 120 time s Figure A 4 Open loop data for z 0 4 30 Valve opening 0 16 0 14 0 12 0 1 0 08 9 98 20 40 60 80 100 120 time s P1 barg P2 barg O e i o gt P

Download Pdf Manuals

Related Search

Abstract abstract abstract art abstract definition abstract expressionism abstract wallpaper abstract synonym abstract of judgment abstract thinking abstract paintings abstract noun abstract examples abstract background abstract ideas abstract class abstract reasoning abstract art images abstract coloring pages abstract fonts abstract class c# abstract chain abstract logo abstracto abstract method abstract class java abstract daddy

Related Contents

Manuel d`utilisation  la feuille d`infos du cira #58janvier 2005 - CIRA-Marseille  Samsung i80 คู่มือการใช้งาน  Kenroy Home 32264GBRZ Installation Guide    Frigidaire FPGC3087MS Installation Instructions  Distribution Boxes, Power Adapters, Hidden Cameras  THE BASICS OF ROBOTICS    Bedienungsanleitung Aritco 7000er Serie  

Copyright © All rights reserved. Failed to retrieve file