to Model Documentation
Contents
1. Variable Valves As with the naturally aspirated engine electronic valve operation and other variable valve mechanisms like dual cams could be investigated using the detailed model Similarly the engine itself could be easily modified to represent a diesel engine or an Atkinson Miller engine for hybrid cars Intercooler For both the detailed and the system level model a more realistic intercooler should be added including thermal lags that are associated with a significant mass of structural material and perhaps water if a water cooled design is used Control Valves More detailed waste gate and Pop off models including improved controls could be added This can be as elaborate as time dependent valve stem or gate motions VGT Variable geometry turbines VGTs in which the nozzle vanes are controlled to both improve performance and to eliminate the need for a waste gate is an interesting extension along with the requisite control system In SINDA FLUINT this is accomplished by a using family of turbines in parallel one for each vane position interpolating between these fixed map turbines based on the current vane position This technique has also been used for variable blade angles in other types of turbines Alternatives to this interpolated approach include dynamically adjusting a single map using expressions and or user logic or perhaps avoiding maps altogether by using a concurrently executed meanline turbine program e2. 2 4265 2 40e5 Pop off Flaw 2 0080 Int Man Press 2 35665 2 3465 2 3265 2 3065 2 2865 2 25065 2 2465 2 22805 2 2065 2 18e5 2 15e5 2 1465 2 1265 2 10e5 2 08e5 2 06e5 2 0465 2 0265 2 00e5 1 98e5 1 96e5 1 94e5 1 92e5 1 90e5 Mass Flow Rate kg s Ed 5946 3405584d Time sec Unfortunately the damage has been done Residual inertia in the shaft plus residual pressure and temperature in the intake manifold perhaps due to undersizing the pop off valve and or intercooler causes the compressor to go into surge at 9 seconds This happens 3 seconds after the engine began to slow down as shown below The turbine LIMP flag shows that it is moving out of its given map range and that extrapolations are being used which are not much of a concern as is surge in the compressor Turbomachine Status During Engine Speed Transient Compressor 0 2 me Turbine g 1 2 3 4 7 6 Fi 8 g 10 11 12 Time sec As was mentioned before compressor surge can be detected not simulated with confidence so anything past 9 seconds should be disregarded Instead the control system and or the compressor should be redesigned to avoid this situation Nonetheless it is interesting that the surge continues to happen long after the intake pressure drops and the pop off valve has nearly closed again Possible Model Extensions This section briefly describes further work or model extensions that are possible or even advisable
3. This software is offered by Turbomachinery Aerodynamic Not all of the turbomachines described have been placed into the Sinaps library For example it only contains low inlet temperature turbines However the raw data is available upon request and it only takes a few minutes to cut and paste performance maps from an Excel sheet into the Sinaps tables Surge lines can be identified separately if they do not exactly correspond to the edge of the map Technology in Greensburg Pennsylvania www turb aero com The proprietor is Ronald Aungier Ron s software enables representative design concepts to be generated and evaluated quickly pe mn i h ITUN Cut away Turbocharger with Color coded Temperature Zones Wikipedia org Compressor Design In total five compressors were designed to support the modeling needs Each was a single centrifugal stage with a basic volute type outlet Initial estimates of the ingested air flow rate of 0 145 kg s 0 32 lb s in non turbo charged mode were used as a starting point neglecting the boost in volumetric efficiency This formed the low flow rate designs Expecting a boosted intake pressure of twice atmospheric 2 bar more compressors were designed for 0 29 kg s 0 64 Ib s Finally medium flow rate designs were added to allow additional pressure losses in air filters mufflers etc Compressors were for the most part designed in pairs with one compressor in a pair ha
4. solve for speedTC call destab S summarize the solver s in the output file call diftfegli 1 speedTC 2 pi 60 0 0 S initialize the ODE timend 12 0 set the end time to 12 seconds call transient S run the engine speed transient At an engine speed of 3500 rpm the Solver finds the balance point to be about speedTC 57000 rpm 888 NERVUS is also set to 3 to avoid halting for nonconvergence which can happen due to compressor surge stall and more rarely due to valve open close instabilities in off design initial conditions The turbocharger s response to the transient is shown below including a reprise of the engine speed variation during the 12 second event Engine Speed Transient 6 000 5 900 5 800 2 700 Engine 5 500 5 500 5 400 5 200 5 200 5 100 5 000 4 900 4 800 4 700 4 600 4 500 4 400 4 300 4 200 4 100 4 000 3 900 3 800 3 700 3 600 3 500 3 400 3 300 3 200 3 100 3 000 7 20e4 Turbocharger 7 10e4 7 00e4 6 90e4 6 80e4 6 70e4 5 00e4 6 50e4 6 40e4 6 30e4 Turbocharger Speed mm fluds paads auibu3 6 20e4 6 10e4 6 00e4 5 90e4 5 80e4 5 70e4 005 1 1 5 2 2 5 3 3 5 445 5 5 5 6 6 5 7 7 5 8 8 5 9 9 5 1010 511 11 512 Time sec The boost lags by a second or two but even more interesting events happen during the slow down portion of the transient as will be described below The torque response below shows that the turbine torque ramps up qui
5. is shown below Turbine Exhaust Manifold LL Ga CDan S F n Co a OD tina ef aa em o Waste Gate oe P Muffler Catalytic Converter Exhaust Runners 1 Gia Exhaust Valves Cylinders Intake Valves Shaft ODE Logic here Intake and Filter rd Mass Flow Sensor czy Throttle ps a p Intercooler Compressor me ms mm ry aame Pop off Valve Sinaps Schematic with cylinder heat transfer on an invisible layer Comparison with the non turbocharged model reveals new components in the intake system compressor intercooler and pop off valve and in the exhaust system turbine and waste gate The mechanical connection between the turbine and compressor is behind the scenes mathematically so these two components need not be shown adjacent to each other In the detailed model see the network logic in the plenum air 1000 as indicated by the comment Shaft ODE Logic here The intercooler is a highly oversimplified HTNS segment oriented convection tie between the compressor outlet and the duct representing the intercooler As noted above this place holder component only removes a few degrees of compressor superheat and does not contribute much pressure drop either The baseline compressor is Compressor 5 and the baseline turbine is Turbine 6 described above These are paths 125 and 225 respectively so net_torque air torq125 air torq225 Alternate turbines and compr
6. rom case is actually bordering on unacceptable A starting speed of 3500 rpm was there used in the system transient to avoid compressor surge as an initial condition Net Torque vs speed at several engine RPM 2000 rpm 3000 rpm 4000 rpm 3 5 4 6000 rpm 1 5 cc Net Torque N m I Lu 3 00e4 3 50e4 4 00e4 4 50e4 5 00e4 5 50e4 6 00e4 6 50e4 7 00e4 7 50e4 8 00e4 Turbocharger Speed rpm The 6000 rpm case appears to be stable at a turbocharger speed of about 72000 rpm However it should be noted that the pop off valve starts to open past 75000 rpm at this engine speed under steady conditions Engine Speed Transient It is desired to impose the following engine speed profile on the system level model and see how the turbocharger responds Imposed Engine Speed 6500 6400 6300 6200 6100 6000 3900 3800 23 00 3600 3500 3400 3300 3200 3100 3000 4900 4800 4700 4600 4500 4400 4300 4200 4100 4000 3900 38600 3700 3600 3500 Engine Speed rpm 0 0 5 1 15 2 245 3 35 4 45 5 55 6 65 7 75 amp 8 5 9 9 5 10 10 5 11 11 5 12 Time sec To achieve this profile with the throttle is assumed to stay wide open the engine load is assumed to vary as needed The system is initially at steady state at soeed 3500 rpm The engine then ramps up linearly to 6500 rpm over 3 seconds holds at that speed for 3 more seconds then it ramps back down to 3500 rpm within another 3 seconds The above pr
7. waste gate is critical since the torques will balance at a partially opened waste gate position see the steep portion of the net torque versus turbocharger speed plots in the system level model below and see the comment on turbine 10 oversizing above If a real waste gate were modeled including its complete dynamic response the solution at the balance point might correspond to a time averaged position or flow resistance A bang bang control system without any realistic lag would have no steady state solution for the system level model and would disrupt cyclic convergence for the detailed level model Even in a transient model of a realistic controller proportional control must be present to generate useful initial conditions for the transient event and it can be disabled during the transient Using the SINDA FLUINT NSOL flag to detect the type of the current solution a slight modification of the valve logic that accomplishes this change from proportional to bang bang control is as follows c so not too wild in steady states use proportional control elseif NSOL eq O then aori this pl up Ppop close Ppop Open Ppop Close Apop_Open AF this endif For a more realistic model of a control valve see the refrigeration cycle TXV example in the SINDA FLUINT sample problem manual and in the Sinaps sample problem set Detailed Model The network diagram for the detailed model colored by lump node temperature and path flow rate
8. 0 71 4 0 7 4 0 69 4 0 68 4 0 67 0 66 0 65 0 64 0 63 Speed 50000 Speed 60000 Speed 70000 Speed 80000 F Ratio Performance Map of Flows for Turbine 6 Speed 50000 Speed 60000 Speed 70000 Speed 80000 Pressure Ratio Performance Map of Efficiencies for Turbine 6 Summary Tables for Turbomachines Low s o toa o s o ore fe ae we DS eh Low alternate Turbomachinery Dimensions For simplicity all turbines were assumed to have the same dimensions a blade tip diameter DTURB of 9 4cm 3 7 an inlet flow area of 1 85e 3 m 1 99e 2 ft and an outlet flow area of 2 13e 3 m 2 29e 2 ft This corresponds to an inlet diameter of 4 9cm 1 91 and an outlet diameter of 5 2cm 2 05 Similarly all compressors had the same outlet flow area of 1 36e 3 m 1 46e 2 ft corresponding to a diameter of 4 2cm 1 64 The low flow compressors had an inlet flow area of 1 46e 3 m 1 57e 2 ft corresponding to a diameter of 4 3cm 1 70 However for medium and higher flow compressors the inlet was raised by 30 14 greater diameter to avoid high Mach numbers at high engine speeds When kinetic energies are significant flow areas should be continuous at a static STAT NORM lump if the two adjoining flow areas are not equal kinetic energy can be gained or lost If a real change in flow area occurs it should happen acr
9. A a EEIE LTR EATE TN 3 ia KE a i N 0 25 AAEE SS Die PAENNEEN 5 ae ae ar cel Elias CEEC Deia pp me Pe ei ersari a P N ae see a A R22 A E E A i a a em ial a ae ae o i Dt a aa e S a a E E o Cail ee E E EE EE 5 a a APEN 0 02 OES re eae pore DU cercat eeteteteo atentament RE eae ees 5 a e El men A Diinan la a m f4 eset point fcvinder design manifolds results fsolution control peontrols The network based logic for the pop off valve in FLOGIC 0 is shown below if pl up le Ppop Close then aori this 0 0 elseif pl up ge Ppop_Open then aori this Apop_Open AF this c so not too wild use proportional control instead of bang bang else aori this pl up Ppop close Ppop Open Ppop Close Apop_Open AF this endif The logic for the waste gate is similar except that since it controls a remote pressure pl up is replaced by the lump ID for the intake manifold pl150 Why proportional control and not just bang bang control open closed with a deadband or on off control which can be achieved by commenting out the last ELSE subblock in either the waste gate or pop off valve logic The answer is primarily a matter of numerical convenience Lacking more realistic valve response and control information and remembering that these elements are merely place holders for a more design specific model no effort was expended to model a more realistic control scheme The response of the
10. C is zero either by parametric sweeps or by using the SINDA FLUINT Solver in a goal seek mode This method can be used to either verify system sizing or for producing the initial condition needed to start a transient In a transient a first order differential equation T I da dt can be co solved using the ODE utilities in SINDA FLUINT ODE 1 is initialized in OPERATIONS before a transient call Grziregia l speedTe 2 61760 0 0 and is updated in FLOGIC 2 at the end of every time step Call diffegl l stest inertia friction 2 pPi 60s Ret torque speedTC stest 2 pix 60 Note that the ODE is written in terms of radians per second requiring conversion to and from revolutions per second stest being a temporary variable containing the speed in rad sec The system level model permits one such hidden area change at the engine resulting in a warning of DISCONTINUITY IN FLOW AREA DETECTED at the junction representing engine heating where energy is imposed as a boundary condition so it need not be conserved Modifications to the Original Engine As was mentioned above the detailed model was derived from a naturally aspirated non turbocharged engine model This section describes the changes made to that model Intake and Exhaust System The intake runners were enlarged slightly from 45mm to 50mm diameter but their length was greatly shortened from 500mm to 150mm given that tuning was no longer required in a tu
11. Compressor 5 Turbine Design Before the turbine components are described in detail a word about the turbine design philosophy is in order Turbines were designed in pairs to support particular compressor component units The difference between the two turbines in a pair e g Turbines 1 and 2 constitute a pair is the expected turbine inlet temperature Due to uncertainties in the expected inlet temperature in this application a high and low inlet temperature turbine was designed for each compressor The high inlet temp turbine in a pair was designed for an anticipated turbine inlet temperature of 1220K 2200 R The low inlet temp turbine in the pair was designed for a 695K 1250 R inlet temperature Each turbine of the pair was designed to be driven by the air mass flow rate that was supplied by the particular compressor that was being supported In general the design problem for each turbine was to find the turbine inlet pressure to be supplied by the IC engine to generate the required turbine power The turbine exit static pressure was fixed at 105 kPa 15 2 psia for all turbines except 5 and 6 These two units had their exit Static pressures fixed at 124 kPa 18 psia in accordance with data from previous IC engine model runs The design power level for each turbine was set at 110 of the design point power for the compressor being supported A 10 power surplus was enforced at the design operating point in ord
12. INDA FLUINT produces to communicate the turbomachine Status For a variable displacement compressor COMPRESS device LIMP 0 nominal within range LIMP 11 or 12 speed is too low or too high respectively compared to input map values pressures and flows are extrapolated but not efficiencies LIMP 1 compressor is choking LIMIP 2 compressor is surging Turbine and Compressor Status Flags Last Cam Cycle at 3000 rpm 0 0 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 1 0 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 2 0 Turbine Compressor normal 2 surge LIMP 0 0 125 0 13 0 135 0 14 0 145 0 15 0 155 Time Letting the shaft speed increase would only make this problem worse Instead the waste gate control or compressor design would need more attention first Finally note that this case was run assuming a fully open throttle Like most programs SINDA FLUINT cannot actually simulate a compressor during stall or recovery if such behavior is not already contained in the map and that is almost never true given the nature of the surge event Instead SINDA FLUINT simply assumes that the edges of the valid map can be extrapolated into nearly flat curves of pressure ratio versus flow rate that the compressor spins without making more significant progress in flow as the outlet pressure increases This is evident in the flat spots at the bottom of the compressor efficiency since efficie
13. The displacement of this compressor is the full engine displacement divided by 60 to allow speed to still be specified in rev min rather than rev sec and the speed of the compressor is the engine speed divided by 2 being a four cycle engine Dividing the displacement by the same factor of 2 would have achieved the same result The combustion heating is replaced by a heating rate that is assumed proportional to the engine mass flow rate This is achieved by applying the imposed QL heating rate in Watts on the engine outlet junction 1 This heating rate is determined by a curve fit based on exhaust temperature heating predicted from the detailed level model based on a second order polynomial Oexhaust Qe0 Qel air FRI Qe2tair FRI 2 where air FR1 is the mass flow rate in kg s of the engine as represented by the positive displacement compression path 1 The work that this compressor adds has already been considered in the effective exhaust heating in the detailed level model so it is subtracted for the net junction heating term OL Qexhaust QTMtthis where QTM this is the net turbomachinery work imposed on this junction engine exit point This system level engine representation is not intended to be either a template or a recommendation It was merely convenient and since the focus of this model is on the turbocharger further refinement was not deemed necessary See also the closing notes on a
14. Turbocharged IC Engine Model This document describes two Sinaps based SINDA FLUINT models of a turbocharged four cycle inline six cylinder 3 7L gasoline internal combustion IC engine The purpose of these models is to demonstrate modeling capabilities and techniques The models are intended to be representative of a typical design but are otherwise notional They are intended to provide starting points for a user requiring a more detailed analysis of an actual engine design The reason why more than one model was created is that there are many areas of interest for such integrated turbocharger engine simulations from piston induced pressure waves to boost lag during acceleration Therefore both a detailed and a system level model were produced to allow focus on either short or long time scale events respectively The first model focuses on the very short time scale events lt 1 second including pressure waves in the intake and exhaust runners and their effect on turbomachinery speed and performance This detailed model could be extended to investigations of waste gate controls EGR Exhaust Gas Recirculation transients etc The second model focuses on longer time scale events gt 1 second up to oo steady state This system level model is appropriate for steady state sizing or for simulating vehicle level transients such as boost lag or pop off valve actuation during throttle variations In this system level model an effective t
15. anifold actually drop and as will be shown later the waste gate closes This is caused by the boost lag the engine is depressurizing the intake system faster than the compressor can replenish it P and Tin Intake Manifold during Engine Transient 2 4565 155 Pressure Temperature 2 4085 150 2 3565 145 2 30e5 140 a 2 9 9585 135 3 s T Ma o p 220e5 130 E m T ta w 2 1585 125 E a 2 10e5 120 2 05e5 115 2 00e5 110 1 95e5 f 105 1 90e5 100 005 115225 3 3 5 4 4 5 5 5 5 66 5 7 7 5 8 B 5 9 9 5 1010 511 11 512 Time sec The turbine event at 3 5 seconds is hardly felt by the compressor but it feels the second event at 6 25 seconds These events are of course the opening of the vvaste gate and the pop off valve At 3500 rpm the vvaste gate is partially open After it closes during the initial lag event by 3 5 seconds the intake manifold pressure has risen again beyond its 2 0 bar trigger point and the waste gate begins to open again this time in earnest Flow is suddenly diverted from the turbine dropping its torque However the shaft speed continues to rise and so the compressor continues to raise the intake manifold pressure to the next threshold the opening of the pop off valve at 2 2 bar These flows are shown below with the manifold pressure reprised for reference Waste Gate and Pop off Control Responses to Engine Transient 2 4865 2 46e5 7 4405 Turbine Flow WG Flow Compr Flow
16. ckly as the engine speed up so the lag in the above speed curve is due to shaft inertia primarily and not due as much to the pressurization and heating of the exhaust manifold At about 3 5 seconds the turbine torque drops suddenly and starts to level off But it drops sharply again at about 6 25 seconds At the end of the 12 second transient with the engine back to 3500 rpm though only for 3 seconds the compressor torque is higher than the steady initial value while the turbine torque is less than the steady initial value However the turbocharger speed is back down to nearly the steady value This means that if the transient had continued the turbocharger would have overshot the equilibrium point and gone to speeds well below 57000 rpm and would have probably oscillated about the equilibrium point after experiencing this disturbance As will be shown later the compressor is actually surging after about 11 5 seconds either the control design for this engine is defective or the compressor design needs to be revisited Torque Response During Engine Speed Transient 9 00 8 50 000 Se a ee Gs ee ee a _Net Torque 7 00 l 6 50 6 00 5 50 5 00 4 50 4 00 3 50 3 00 2 50 2 00 1 50 1 00 0 50 N m Torque Time sec The pressures and temperatures in the intake manifold are shown below Notice that in first few seconds the temperature and pressure in the intake m
17. e exhaust system the unmodified exhaust system is not well tuned for the turbocharged engine While the waste gate remains closed at 3000 rpm it is partially open at 4000 rpm near the steady operating point of the turbocharger as shown in the flow rate plots below Note that the pop off valve is still closed under steady operating conditions at this speed Mass Flow Rates for Last Cam Cycle 4000 rpm Turbine Compressor Pop Off Waste Gate Mass Flow Rate legis 0 125 0 13 0 135 0 14 0 145 Time sec What are steady operating conditions What turbocharger speed speedTC represents a balance point between turbine and compressor torque and by pass flow in the waste gate Unfortunately in this detailed level model with the flow rate in the engine calculated based on piston and valve responses integrated over many crankshaft cycles no steady state solution exists One alternative would be to run the model for a long time starting with a good guess at speedTC or perhaps using artificially decreased inertia to try to arrive at a quasi steady answer without excessive run times Another alternative is a system level model presented later At the steady state balance point with a little higher turbocharger speed than the transient case uses the waste gate is actually slightly open at 3000 rpm It is the system level model which predicts the starting point speedTC for the short time scale results such as shown b
18. elow back to the 3000 rpm case in a plot of torques Engine Response Last Cam Cycle at 3000rpm 55979 55978 55977 55976 55975 55974 55973 55972 55971 55970 55969 55968 55967 55966 55965 55964 55963 55962 55961 55960 55959 55958 55957 55956 55955 55954 55953 55952 55951 55950 55949 55948 Torque N m pwdi pesades 5L 0 354 0 125 0 13 0 135 0 14 0 145 0 15 0 155 Time sec The oscillation in the compressor torque is minimal due to the large manifold and short runners in the intake system But the oscillation in turbine torque is large as pulses from each piston arrive and as the temperature swings by almost 30 C at 3000 rpm which is likely an overestimate due to the assumption of adiabatic ducting This variation in turbine inlet conditions causes a large oscillation in the net torque on the turbocharger shaft though the oscillation in the shaft speed is small due to its inertia The trend in the shaft speed is toward an increased speed because the time averaged net torque is positive at about 0 6 N m in this case Why Because the initial guess at speedTC was too low The system level model predicts an equilibrium point of about 57000 rpm If this is the case why not start at an even higher shaft speed The reason is because the compressor is actually transiently surging at these conditions in the 3000 rpm case but not at 4000 and 6000 rpm This is shown by the LIMP indicator flag that S
19. er to require the turbine waste gate to be in some position other than fully closed This provides the turbine compressor power cycle with some nominal control authority Note that when a turbine was designed to support either a vaned or vaneless diffuser compressor compressors also tended to be designed in pairs the turbine performance was designed to support the least efficient compressor of the pair at the design operating point Note that all turbines were radial inflow turbines Turbine 1 Turbine 1 was the high inlet temperature turbine of the two turbine units designed to support compressor 1 At the design speed of 65000 rpm turbine 1 required an inlet pressure of 149 kPa 22 psia total This resulted in a design point pressure ratio of 1 45 Isentropic efficiency at this operating point was approximately 82 Total to Static T S Turbine 2 Turbine 2 was the low inlet temperature turbine of the two turbine units that supported compressor 1 At 65000 rpm turbine 2 required an inlet total pressure of 203 kPa 29 5 psia at 695K 1250 R inlet temperature to generate the required power for a design point pressure ratio of 1 94 T S Isentropic efficiency at this point was essentially identical to turbine 1 Turbine 3 Turbine 3 was the high temperature turbine of the pair that supported compressors 2 and 3 the high flow compressors At 1220K 2200 R inlet temperature 65000 rpm and design point
20. essors can be input or imported from the supplied library The pop off valve flows from the intake manifold to the inlet of the compressor and is controlled by network based logic in FLOGIC O start of time step as documented above The waste gate by passes the turbine and uses similar logic to the pop off valve to perform its control function as described above Sample Results This section provides examples of the types of responses that are possible to explore for the detailed level short time scale model The pressures in the manifolds and runners next to the valve are plotted below for the 3000 rpm case Pressure Response for Last Cam Cycle 3000 rpm 2 60e5 Exh Valve 2 55e5 Int Valve Exh Manifold Int Manifold 2 0065 2 4565 2 40e5 2 3565 2 3065 2 2565 2 2065 Pressure fabs Paj 2 15e5 2 10e5 2 0565 2 00e5 1 95e5 1 90e5 0 125 0 13 0 135 0 14 0 145 0 15 0 155 Time sec As expected and by design the variation in the pressure in the intake system is minimal Note that the intake manifold is low enough that both the waste gate and the pop off valve are always closed during this run The exhaust ducting exhibits strong pressure variations Long runners and an effort to minimize losses to preserve a high total pressure at the turbine inlet are responsible for these variations At 6000 rpm these pressure waves are too disruptive and they lower the sweeping effect in th
21. flow rate of 0 29 kg s 0 64 Ib s a turbine inlet total pressure of 176 kPa 25 5 psia was required to generate the needed power This implies a design point pressure ratio of 1 68 Isentropic efficiency at this condition was approximately 75 Turbine 4 Alternate Turbine 4 supported compressors 2 and 3 at an inlet temperature of 695K 1250 R A turbine inlet total pressure of 248 kPa 36 psia was required This resulted in a design point pressure ratio of 2 37 and an isentropic efficiency of approximately 82 Turbine 5 Turbine 5 supported compressors 4 and 5 the medium flow compressors at an inlet temperature of 1220K 2200 R and a design flow of 0 245 kg s 0 54 Ibm s A turbine inlet total pressure of 210 kPa 30 5 psia was required giving a design point pressure ratio of 1 69 at a speed of 75000 rpm Isentropic efficiency was approximately 77 Turbine 6 Baseline Turbine 6 supported compressors 4 and 5 the medium flow compressors at an inlet temperature of 695K 1250 R A turbine inlet total pressure of 324 kPa 47 psia was required giving a design point pressure ratio of 2 61 at a speed of 70000 rpm Isentropic efficiency was approximately 82 Flow lbm sec Efficiency 0 51 0 54 0 49 4 0 48 4 0 47 0 46 0 45 0 44 0 43 0 42 4 0 41 0 4 0 39 0 38 0 37 4 0 36 0 35 0 83 4 0 82 4 0 81 0 8 0 79 4 0 78 0 77 0 76 4 0 75 4 0 74 0 73 0 72 4
22. fluid model The design point was chosen as 4000 rpm with intake at sea level and 25 C It was desired to evaluate performance over the range of 2000 to 6000 rpm One of the main reasons for an engine turbocharger model is to verify the design of the combined system Therefore it may be no surprise that several iterations were performed on the turbine and compressor design before a satisfactory pair was chosen It was decided to make all variations that were explored available to the user in the form of a Sinaps turbomachine library file tm_library xml Therefore a total of 5 compressors and 6 turbines are described below COMPRESSOR ame SECTION HUE WA COMPRESSOR i MUL SIs TURKE EXHAUST DAS CAJT LET TEL OW as mer TURBINE AIR DISCHARGE SECTION COMPAS SOR AMBIENT AIR COMPRE SS OR IMLET WHEEL Turbocharger Wikipedia org One way to evaluate a turbomachine within SINDA FLUINT is to provide performance maps For compressors a map of pressure ratio vs mass flow rate for several speed lines plus another map of isentropic efficiency vs mass flow for several speed lines will satisfy this need The edges of the pressure flow map are assumed to represent choking and surge lines Similar maps are needed for turbines as described in detail in the SINDA FLUINT manual The design concepts for all turbomachines presented herein were generated with the software packages known as CompAero and TurbAero
23. g TurbAero to serve the current operating point at each time step or iteration Integrated Meanline Programs Extending the simulation to handle VGTs is actually just one reason to consider the integration of the models with meanline software that predicts turbine and compressor operating points e g CompAero and TurbAero A more urgent need is a combined design tool which would help reduce the manual iterations needed to find and to verify a suitable turbocharger design for a given engine design and a given set of operational scenarios In this mode the actual turbine or compressor design itself might vary in between iterations or between steady state or transient solutions perhaps as driven by SINDA FLUINT Advanced Design Modules such as the Solver and the Reliability Engineering module Multiple Compressors and Turbines Expanding to two turbochargers whether in parallel perhaps for packaging or in series perhaps to expand the range of boosted engine speeds to lower RPM or to increase the boost at higher RPM would be relatively trivial 7 to accomplish using the techniques demonstrated here Any number of shaft speeds can be solved simultaneously whether in steady state balancing using the Solver or in transients using multiple co solved ODEs EGR Other system level transients and interactions could be explored including EGR exhaust gas recirculation systems for knock avoidance and emissions control perhaps integ
24. gn point is 66 T T When operated with the IC engine Sinaps model it was discovered that the design flow rate of 0 145 kg s 0 32 Ibm s was insufficient to support engine operation at the boost levels provided The system model portrayed the compressor as operating perpetually in a choked state which is physically consistent with what would be expected under these circumstances Compressor 2 This compressor was sized to produce a mass flow of 0 29 kg s 0 64 Ib s at approximately the same pressure ratio as compressor 1 Compressor 2 retained the same design speed 65000 rpm as compressor 1 Compressor 2 s impeller tip soeed was set at 427 m s 1400 ft s identical to Compressor 1 It was also equipped with a vaned diffuser to improve design point efficiency Design point efficiency is approximately 77 T T The improvement over Compressor 1 is largely due to the increased mass flow rate When operated with the IC engine model compressor 2 experienced operation to the left of the surge line Compressor 3 Alternate This compressor was sized to the same design point mass flow rate and pressure ratio as compressor 2 The design operating speed was also the same as compressor 2 However a vaneless diffuser was substituted for the vaned diffuser of compressor 2 The vaneless diffuser resulted in a drop in design point efficiency to approximately 75 but provided an increased stable flow range Like compressor 2 3 experienced su
25. hybrid detailed system model These values are calculated using the Excel Solver and an RMS fit method with double weighting applied to the design point of 4000 rpm Parametric Sweeps Having achieved a model that is capable of solving for inexpensive steady state solutions many such solutions can be explored at once using a parametric sweep While any key variable can be swept in this case a lot of information can be gained by exploring results as a function of turbocharger speed speedTC For example below is a sweep of flow rates through the turbine and waste gate plus pressure ratios across the turbomachines for the 4000 rpm engine speed case with a wide open throttle WOT Flow and Pressure Ratio vs speed at 4000 engine RPM 0 26 0 255 0 25 0 245 0 24 0 235 0 23 0 225 O 22 0 215 0 21 0 205 0 2 0 195 0 19 0 185 0 18 0 175 0 17 0 165 0 16 0 155 0 15 4 0 145 0 14 0 135 0 13 0 125 0 12 0 115 O 11 FR Turb Pratio COMP Pratio TURB FR Comp Mass Flow Rate flag s Oey 8155814 4 00e4 4 50e4 5 00e4 5 50e4 6 00e4 6 50e4 7 00e4 7 50e4 8 00e4 Turbocharger Speed rpm The waste gate opens at about 58000 rpm and is fully open by 63000 rpm successfully dropping the turbine pressure ratio as a result of its by pass action A plot of the net shaft torque net_torque during the same sweep shows that the stable operating speed of the turbocharger is about 61000 rom where com
26. ication see below The volumes within the intake upstream of the air filter have also been neglected using FLUINT junctions instead of tanks as has the inertia of the ducts at the larger time scales that this model operates using FLUINT STUBEs instead of tubes The baseline compressor is still Compressor 5 and the baseline turbine is still Turbine 6 described above These are paths 3 and 4 respectively so net_torque air torq3 air torg4 As with the detailed model alternate turbines and compressors can be input or imported from the supplied library The boundary node labeled speedTC in the middle of the diagram is a postprocessing trick that enables the user to check the current turbocharger speed using hover text The temperature of this node which has no effect on the solution is set to the register soeedTC This trick is not necessary for plotting since registers can be plotted directly Simplified Engine Model It was desired to operate at long enough time scales that the actions of each piston and valve could be averaged into a steady flow rate and heating value The engine flow rate is therefore predicted as a positive displacement compressor using the intake VE values from the detailed model calculated using local densities calculated in vol_eff_sys in the intake manifold versus atmospheric density which is the basis of vol_ eff in the detailed model This VE is imposed as a function of RPM
27. ime averaged engine representation is used to avoid resolving the details associated with resolving each crankshaft revolution but the turbocharger and valve models are identical to those used in the detailed model These models were developed as extensions of a model of a naturally aspirated non turbocharged engine design Only those parts of the design and models that differ will be described here The original model is documented separately http www crtech com applications ICengine html Basic knowledge of SINDA FLUINT modeling is presumed The model is built parametrically using registers user defined parameters whose names are indicated within the text using italics such as Kair_filter Many variations can be made simply by adjusting these values while others will require changes to the thermal fluid network and associated user defined co executing logic The model was built using SI units though some inputs were converted from English US Customary units The temptation to do it all in one model should be avoided A model built for short time scale events would be inefficient to execute for long time scales A steady state model operates at infinite time scales for example and is an important tool for sizing and sensitivity studies Compressor and Turbine Designs This section briefly explains the performance criteria and rationale for the compressor and turbine components for the IC engine turbocharger thermal
28. ncies cannot be safely extrapolated past the boundaries of the map Turbomachine Efficiencies for Last Cam Cycle at 3000rpm 0 8 0 795 0 79 0 785 0 78 0 775 0 77 0 765 0 76 0 755 0 75 0 745 0 74 0 735 0 73 0 725 0 72 0 715 0 71 0 705 0 7 0 695 Turbine Compressor Isentropic Efficiency 0 125 0 13 0 135 0 14 0 145 0 15 0 155 Time sec In reality of course the efficiency should drop precipitously when the compressor surges The problem is more acute at 2000 rpm and at that engine speed turbocharger speed is also well below the available maps at about speedTC 33000 rpm Therefore engine speeds lower than 3000 rpm were not investigated for the turbocharged engine detailed model with a fully open throttle This raises an issue it is difficult to generate reliable performance maps far from the design point To go too far away from that point advanced compressor analysis methods or test data would be needed By the way for a turbine LIMP has the following meaning LIMP 0 nominal within range LIMP 1 or 2 pressure ratio or difference is too low or too high respectively LIMP 1 or 2 flow rate is too low or too high respectively LIMP 11 or 12 speed is too low or too high respectively Is the use of steady state performance maps for the turbine and compressor during a short time scale event such as those depicted above acceptable One answer is yes simply because the alternative
29. ofile is specified as a Sinaps Table which is interpolated in the case named rpm up asa function of time It can therefore be easily altered to follow another profile This Table look up logic is placed in the global FLOGIC O block using a library interpolation routine and the knowledge that the Table is Array 3 and that the current simulation time is timen call dldegl timen engine a3 speed This profile was chosen for model testing purposes and for demonstrating features and does not obviously represent any realistic drive case For the initial condition the speedTC that balances the turbocharger must be found as if the parametric sweeps shown above were run and SINDA FLUINT itself were charged with finding the zero net torque point To accomplish this a SINDA FLUINT Solver run is set up with speedTC as the sole design variable with OBJECT net_torque friction speedTC and GOAL 0 goal seeking as shown in the form below FE Solyer Reliability Data Solver Reliability Gol 0o Nemus 3 tolerate also nonconvergence in E Nloope 100 BETA Oro eta cal to procedue Object net_torgue tiction speedTE Pusho Mo Metho 2 Fletcher Reeves method or Sequer Ficacto Bi Adera 000 Acero 00300 Adero 02 Achga Op Aerie Aehao Ba Rerro Bn Mwrkro P Mdero Pe Mrki P Nconvo BO Revio 20 Cancel f A Use of the Solver to find initial conditions requires a custom OPERATIONS block call solver S find the balance point
30. ons 2 order ODE with frictional hysteresis etc tabulated losses versus valve position etc In fact one of the purposes of these engine models is to enable valve interactions to be investigated However since these models are for demonstration and template purposes only and since actual valve data is missing the usual short cut of modeling a valve as an orifice is employed Furthermore a very simple control system is employed flow area in proportion to the controlled pressure with the pressure of the intake manifold being controlled by both the waste gate and pop off valve also known as blow off valve The waste gate is assumed to be fully open 25 of the turbine outlet flow area or Awg_open 0 25 at an intake manifold pressure of 2 2 bar Pwg_open and fully closed at an intake manifold pressure below 2 0 bar Pvvg close It is assumed to respond instantly and proportionately to any pressure in between Similarly the pop off valve is assumed to be a 2cm 0 8 diameter Dpop proportional valve which is 25 open Apop_open at an intake manifold pressure of 2 4 bar Ppop open and closed at a pressure of 2 2 bar Ppop close As the intake pressure rises the waste gate fully opens before the pop off valve begins to open using these particular settings Mame Type Expression Comment value 1 Puig Open Float 2 e5 waste gate crack pressure Compressor outlet 220000 0 5 ies tiene n ie e tantes Dm mi oe ee manera
31. oss a path and not across a lump Therefore the turbomachinery inlet and outlet flow areas are applied to the upstream and downstream paths to make sure that flow area transitions were not hidden from the model but rather were exposed as an intentional flow processes as needed to completely conserve energy Turbocharger Mass and Speed Response Assuming an aluminum compressor disc and a steel shaft and turbine disc the inertia of the rotating portion of the turbocharger was estimated to be 7 68e 3 kg m 5 65e 3 slug ft For convenience this value register inertia was assumed constant for all compressor and turbine combinations though variations of this number are explored below Friction was assumed to be proportional to the turbocharger speed with a value of 1 0e 6 N m rpm register friction Like many other parts of this model this value is arbitrary lacking better data and acts mostly as a demonstration of how to apply friction and a place holder for replacement or update when a design specific value is available Torque is estimated by the turbine and compressor devices and is positive for work done on the fluid system The negative of the sum of these torques stored as the register net_torque represents the net torque on the turbocharger shaft if it were frictionless In a steady state analysis the turbocharger shaft speed speedTC can be varied until the point at which the torque on the shaft net_torque friction speedT
32. pressor and turbine torques are balanced This value can be used as the starting point for the detailed level model above More extensive multi variate design space sampling and searching methods are available as well Net Torque vs speed at 4000 engine RPM 1 5 0 5 I a En Net Torque N m I Ln 2 5 3 5 4 00e4 4 50e4 5 00e4 5 aed 6 00e4 6 50e4 7 O0e4 7 00e4 8 00e4 Turbocharger Speed rpm Note the steep slope at the balance point this corresponds to the active region of the waste gate as intended by the turbine design This steep slope also means that speed variations will be minimal if the engine operates at this point the waste gate will regulate soeed Outside of this active control range more rapid shaft speed changes can be expected The plot below is the same information but including the 2000 3000 and 6000 rpm engine speed cases as well The balance point at 3000 rpm is not much different than at 4000 rpm but the 2000 rpm balance point is a very slow 33000 rpm However this data must be disregarded a plot of the compressor LIMP factor shows the compressor is surging at nearly all points along the 2000 rpm line and a re design of either the compressor or the waste gate controller would be needed to extend the operating range to such slow speeds This was previously discussed in the detailed level model Once again as a result only speeds higher than 3000 rpm were investigated The 3000
33. rated with Tflame adjustments if a more complete combustion model were not also used EGR systems can have strong and often adverse interactions with turbocharger design and operation Hybrid Detailed System Model The system level model could not predict engine flows and exit temperatures and the detailed level model is not efficient for predicting turbocharger balance points which it often needs as an initial condition These models were developed in parallel passing data back and forth between them In essence the system level model was calibrated against the detailed level model A careful reviewer will note that these iterations were not completely closed Both models could be available as submodels within a single Sinaps model automating the passing of data and iterative closure Submodels can be turned on or off dynamically within a single run as needed Rev 1 October 10 2011 www crtech com we Complete with extra check valves and other by pass controls
34. rbocharged system The box shaped intake manifold was replaced with an extension of the intercooler a 61cm 2 Lintake_man long by 10cm 4 Dintake_man diameter pipe This manifold like the box shaped one in the original engine model was again assumed to be stagnant and was not resolved axially it was assumed to operate as a perfect manifold with a single pressure given the small L D ratio The intake duct became the intercooler see below so a new intake section 60cm long by 10cm diameter was introduced with a reduction in Kair_filter from 20 to 10 to keep the air filter pressure drop the same despite a nearly twofold increase in flow rate thanks to turbocharging eS 1 Dintercool Float 4 0 0254 diameter of intercooler 0 1016 5 os cs a a a S ee ee ee EC P ama 5 i naa oe ae e A E peri Pot atrets nisi OC A RA El A a i E OEE a E rae 40 0254 a ee ee ell eee eee eee ee Ct al co ee a El 150 3046 E attendees ee natal z Car Float zaoo 0 nn ggg T A Cer ee U El o T E E E e a CO Ma TT uena a A i EE El Ae Tee eee etre ss Dates sign ic Pisen de nt eer een ee e T ice RM a a a ale rama tral ee a a t a EE Lot aletes seh ae oraitmoomntme a i nna T o 190 084 E E ert rt ll i OO TTT El El ecos CE EE DER Ie E a mee ii fiter Bat a pt gg ii i a El SE ca ce cc a o Float Gj ios T PNE E AE A ENIA SETTA n se A EC a OD a See ee a ae a rae El tieta de E ec ipea OO a a a El aieiaa Sl i Cle M 4 Hhset poin
35. rge at many steady state operating points portrayed by the system model Compressor 4 This compressor was sized with a medium design point mass flow rate of 0 245 kg s 0 54 lb s and a higher design point pressure ratio of approximately 2 5 to better target the IC engine s apparent air consumption characteristics The higher pressure ratio was provided by an increase in compressor 4 s design speed to 75000 rpm Compressor 4 also incorporated a vaned diffuser to raise design point efficiency which was approximately 78 Compressor 5 Baseline Compressor 5 was a counterpart of 4 with a vaneless diffuser to extend the stable flow range Compressor 5 s design point efficiency was approximately 72 T T due to both the vaneless diffuser and a reduction in design speed from 75000 to 70000 rpm as needed to forestall the occurrence of inlet choking In addition the design inlet condition for Compressor 5 was changed to an inlet stagnation pressure of 86 kPa 12 5 psia in order to better account for the presence of an air filter 0 73 0 72 0 71 0 7 Speed 50000 Flow 0 69 Speed 50000 Eff 0 68 Speed 60000 Flow 0 67 Speed 60000 Eff 0 66 Speed 70000 Flow 0 65 Speed 70000 Eff 0 64 Speed 80000 Flow 0 63 Speed 80000 Eff 0 62 0 61 0 6 0 59 0 58 0 57 0 56 0 55 g 0 54 0 53 0 52 0 51 0 5 0 49 Pressure Ratio ADU BISI4S 025 0 3 0 35 0 4 O45 0 5 0 55 0 6 0 65 Flow lbm s Performance Map for
36. s are virtually unavailable for any system level investigation transient aerodynamics using a CFD program In fact at steady state the 3000 rpm engine speed case continues to experience compressor surge To avoid this as an initial condition the system level acceleration transient described below starts at an engine speed of 3500 rom not being feasible in such a study But with high velocities and high blade speeds the transit times for fluid particles though either of the turbomachines are fortunately short compared to the even the small time scales applied here Recovery from a surge within these time scales is a different matter but as noted above simulating surge itself is not realistic all that can really be accomplished is to flag the potential for it happening and then alter the design to avoid it And that is a valid purpose for this type of model Overall Performance Metrics In the non turbocharged engine model overall performance metrics were calculated at three engine speeds These corresponding calculations in this model yield the following results i 3000 rpm 4000rpm 6000 rpm Turbocharger Speed 56000rpm 62000 rpm 72500 rpm Volumetric Efficiency 122 130 130 Thermal mar 40 694 42 094 36 7 es a The volumetric efficiency VE register vol eff is now greater than 100 due to turbocharging An alternate VE based on local densities rather than atmospheric is also calculated since it is useful a
37. s an input to the system level model as described below This alternate VE is calculated as vol_eff_sys The power developed by the engine is of course much greater than that of the non turbocharged engine which could only produce 84kW of power at 4000 rpm turbocharging has increased power production by 73 at the design point System level Model The inability of the detailed piston valve model to achieve a steady state other than a cyclically repeating pattern was noted above and yet steady states form the basis for many important design studies Furthermore to investigate boost lag and operation of the pop off valve longer time scale transients say 1 to 100 seconds are needed than are feasible using a detailed engine model that resolves each crankshaft rotation into many time steps Therefore an alternative model was developed to allow steady states to be solved and to allow investigation of longer term transient events such as engine speed variations This model is depicted below Lis Exhaust Pipe Exhaust Cair 250 Pe sim Manifold S 1 Muffler i Catalytic Conv Waste Gate Engine Compressor air 60 Intake P Manifold Throttle Pop off Valve Intake and Filter Mass Flow Sensor The intake and exhaust system models including pop off valve and waste gate are virtually identical to the detailed model However the detailed responses of the runners are considered negligible in the interest of simplif
38. t fcvinder design manifolds fresults fsolution control fcontrols As with the naturally aspirated engine the throttle is set to be wide open unless otherwise indicated engine speed variations are assumed to be caused by variations in the load The exhaust system was largely the same in both engines except that the irrecoverable losses bends etc in the muffler and exhaust pipe were reduced a little as if the diameter had been slightly increased to avoid another design iteration with the turbine Intercooler The intercooler was simply a 1 22m 4 long extension of the intake manifold that was assumed to have a wall temperature at ambient 25 C This is not intended to represent a real intercooler it is merely a place holder since it achieves only minor cooling and has an unrealistically low pressure drop as a result Since the purpose of the model was not intended to demonstrate a real heat exchanger either air or water cooled no effort was made to be more realistic in this component If need be a detailed level model of a heat exchanger could be built in C amp R Thermal Desktop and FIoCAD Or effectiveness NTU type methods or compact heat exchanger CHX type methods could be applied using SINDA FLUINT routines such as HXMASTER as documented elsewhere including the SINDA FLUINT User s Manual Waste Gate and Pop Off Valve Detail valve modeling is possible including modeling servo pressure lines valve stem moti
39. ving a vaned diffuser and the other a vaneless diffuser The chief difference between these is that the unit with the vaned diffuser will exhibit somewhat higher efficiency in the area of design flow but have a narrower range over which operation is stable Compressors 1 2 and 3 were designed for a boost of 83 97 kPa 12 14 psid at the design point while compressors 4 and 5 were designed for a somewhat higher boost of approximately 124 kPa 18 psid It should also be stated here that in the interest of time and given the demonstration nature of this modeling task the efforts to optimize each unit s aerodynamic design were not particularly strenuous Several percentage points of efficiency could probably be realized with further refinement CompAero and TurbAero input files corresponding to the machines in the library are available upon request Other meanline turbomachinery software is available from other sources and can be used as well providing performance maps can be created from them Compressor 1 This compressor was sized to produce a design point mass flow rate of 0 145 kg s 0 32 Ib s ata pressure ratio of approximately 2 0 Total to Total T T A relatively conservative value of design operating speed of 65000 rpm was chosen Compressor impeller tip speed was set at 427 m s 1400 ft s The compressor was equipped with a vaned diffuser to improve design point efficiency and a conservatively sized volute Efficiency at the desi
Download Pdf Manuals
Related Search