FSAE: Engine Simulation with Wave 1

Introduction:

The engine used in the FSAE competition by Oakland University is a Honda 600cc engine. Knowledge of this engine at Oakland has been growing as it was the chosen engine for these last few years. The Honda engine used on the 2004 and 2003 car is an F4i engine, while in the 2002 car and prior to that the F2 was used. The restrictor in the intake system makes the engine operate very differently to the stock engine. DTA Fast and Haltech controllers were used by the Oakland team to map the restricted engine. 1D engine simulation software packages are becoming very popular in engine design efforts as they reduce the amount of expensive dynamometer testing required and also make development quicker. The Oakland University team started an effort towards engine modeling during the 2004 season. The engine modeling with Wave software donated for this purpose by Ricardo is described in this report.

FSAE engine testing, room lights off, 20 second exposure photo.

1

2

Experimental data

2.1 Flow bench experiments Flow bench testing is a method of determining flow characteristics of engine cylinder heads. As engine performance is highly dependent on engine breathing capability, flow testing of heads and manifolds is used to improve flow characteristics. The tests reported here were done at a steady vacuum of 28” of water below atmospheric pressure for inlet, and 28” of water above atmospheric for exhaust. Flow rates were recorded at 1mm increments, starting with zero lift to check for leakages and maximum lift was 9mm. Flow tests were performed one cylinder at a time by using a bore tube adaptor that connected the head to the flow bench. The bore tube adaptor had the same diameter as the engine bore (67mm), and was slightly longer (or higher) than the bore (72mm). All tests were performed at wide open throttle. Head: The head tested was an available F2 head from a disassembled engine, Original springs were replaced by very soft compression springs to facilitate the rapid adjustment of valve lift. Intake Manifolds: 1. Stock intake manifold from the F4i engine. 2. Formula 2002 intake manifold as used in competition. 3. Formula 2003 intake manifold as used in the 2003 competition. Testing Procedure: Bore tube adaptor assembled to head on Cylinder #1. A temporary bellmouth was built out of clay to the inlet runner, this aids the air to enter the port. Flow test started at zero lift where leakages were checked to be small (they were always zero in our tests). Flow tests at 28” H2O for valve lifts from 1mm to 9mm every 1mm were performed. Then mounted the manifolds listed above and repeated flow tests at all valve lifts Changed flow bench setup from inlet testing (i.e. suction) to exhaust testing (i.e. blowing). Flow tested exhaust port from 1mm to 9mm at 1mm increments. Then mounted the bore tube adaptor to cylinder2. Flow tested exhaust, changed flow bench to inlet testing and repeated all inlet tests as for cylinder 1. Results and Comments: Figure 1 shows that the stock intake hardly restricts the head. The 2002 and 2003 intakes show the dominant effect of the restrictor. Not much variability between 2002 and 2003 intakes, and not much variability between the runners of the 2003 intake. Hence no need to trim the individual cylinders injector timings. The coefficients of discharge were then calculated from the measured flows so that they were available for engine modeling.

2

Flow rates with lift 140.0

120.0

Intake port Cyl1

100.0

Intake port Cyl2 Intake port Cyl3 Flow CFM

Intake port Cyl4 80.0

Stock Intake Cyl1 Stock Intake Cyl2 2002 Intake Cyl1 2002 Intake Cyl2

60.0

2003 Intake Cyl1 2003 Intake Cyl2 2003 Intake Cyl3

40.0

2003 Intake Cyl4

20.0

0.0 0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

10.0

Valve lift mm

Figure 1: Flow curves from flow bench testing

2.2

Flow Bench Calculations

The measurement of CFM on the flow bench was at the pressure and temperature of the meter. That is for inlet valves, it was 28” H20 lower than atmospheric. Conversion to mass flow for comparison to simulation was preferred as this is irrespective of air pressure and temperature. Mass Flow

m kg m  Patm  28" H 2O  0.0254  1000 3  9.81 2   ft m 1min  gr in m s  MassFlow  CFM  0.0283 3    1000 J 5 min ft 60s kg 287  Tmeter  32   273 kgK 9 3

3

Also SCFM conditions are 29.92” Hg and 70F. SCFM  MassFlow

gr kg 1 m3 60 s ft 3     s 1000 gr  s kg min 0.0283m3

3

Determination of Coefficient of Flow Coefficient of flow (referred to as CF in Wave, here also referred to as Cdva) can be calculated for incompressible or compressible flows, both are close at this pressure difference of 28” H20, i.e. approx 7kPa. For incompressible flow, reference IC Engines 2 class notes gr kg MassFlow  1 s 1000 gr Cdva   NoofValves kg Avalve m 2  2  3  P( Pa ) m Note, since treated as incompressible density can be taken as the atmospheric or at the meter! For compressible flow , reference IC Engines 2 class notes and Heywood gr kg MassFlow  1 s 1000 gr  Cdva  1  1 1 1   2 NoofValves  2       2   P P 1 Avalve m 2  Pup  1   v    v      P  RTup  Pup     1    up   Note Pup denotes upstream pressure, units on denominator not shown for clarity. In Mass Flow calculation for the exhaust, since the pressure at the meter is approx 28”H20 above atmospheric a +ve sign is used in the mass flow equation rather than –ve. The Wave BASIC User Manual section 2.4 describes the way flows are calculated. The isentropic velocity is calculated as, for air,  P vis  7 RT0 1     P0

  

0.286

  

where R = gas constant 287 J/kgK for air T0 = upstream total temperature P = downstream static pressure P0 = upstream total pressure The Effective Area Volumetric flowrate Aeff  vis Cd is based on curtain area, and CF is based on valve area. Aeff 4 Aeff Cd  CF  DL D 2

4

2.3

Engine Geometrical information

Engine details were obtained from the engine manual and by measurement. General information such as bore, stroke, valve lifts, valve opening and closing angles were available in the manual. Connecting rod length, valve, port and runner dimensions were physically measured. The F4i heads and cams available were on good working engines and hence the valve lift with crank angle could not directly be measured as the cams are directly acting on the valves and it was not possible to probe them with a dial indicator. The cam lobe was measured instead, as this was reachable with a dial indicator. Valve lift profile was mathematically determined from the measured cam lobe. The valve lift profile was determined by mathematically imposing a flat follower on the measured lobe. The calculated valve lift matched really well with the angles specified in the manual, total error was 3 degrees. 2.4

Dynamometer Engine tests

The Honda 600cc engine was tested on a Clayton water brake dynamometer. Instrumentation was achieved by National Instruments hardware and LabView. The measured parameters were: speed, torque, air temp, intake air volume flow rate and fuel flow. The DTA engine controller GUI supplied the usual engine sensor data. The dyno data obtained at Oakland was also eventually compared to the torque data obtained from the DynoJet test performed during the 2003 competition.

3

Wave modeling

A 600cc engine model was built following the guidelines in the tutorial in the Wave Knowledge Centre. At first default setting were used for all parameters, then measured data was implemented when it was available. The measured geometrical data implemented in our model is  intake and exhaust Coefficient of flows  intake and exhaust valve lifts  Intake and exhaust valve, port and runner geometry. 3.1

Model Validation

Steady state flow bench models were built to simulate the flow bench tests, refer to Figure2. The pressure difference between the two ambient icons was 7kPa, which approximated the 28” of water during the flow bench tests. The mass flow rate determined from these models was compared to the experimental values. Models were modified until agreement was good. Similar steady state flow bench models for the intake system including the restrictor and the exhaust ports were also built, refer to Figure 3. The steady state modeling of the intake system was very important and had huge effects on the overall engine model. Since the restrictor determines the maximum amount of air the engine can breath, its model proved to be very important for the overall engine performance.

5

Bore tube adaptor

Ambient at 100kPa

Frictionless Y junction and port Frictionless ports No bends, angle,or friction

Ambient at 93kPa

Mass Flow Rate Measurement Cd at right of port is the only loss

Cylinder

Figure 2: Steady state flow bench model for the intake port

Portion of intake plenum

Intake Runner

20mm section Nozzle

Diffuser, modeled with taper

Figure 3: Steady state flow bench model for the restrictor and intake port

6

The whole engine model was then modified to have the details determined from the steady state flow bench models. Hence air flow would be modeled as close as we could to the real engine. Refer to Figure 4.

Injectors Intake plenum

Engine cylinder Exhaust collector

Figure 4: FSAE Full engine model

Engine model volumetric efficiency was compared to the measured from the 2003 engine testing, refer to Figure 5. Air flow rate into the engine was measured on the 2003 engine by means of a laminar flow sensor which provided the raw measurement for the experimental volumetric efficiency. 3.2 Simulations of intake / exhaust systems and cam timing Torque was compared from model and experimental measurement obtained from the dyno tests. Model simulations varying intake and exhaust system geometry were then performed. The intake and exhaust runner lengths were varied and exhaust systems with four into two into one configurations were simulated. Figure 6 shows some of the results of these simulations together with the experimental torque measured on the DynoJet test. Simulations varying cam phasing and cam timing were also performed but results are not being shown here. Since the Honda F4i engine is dual camshaft, changing cam phase does not involve much hardware. Coming up with a different cam duration is harder to implement physically, but the theory and building of a cam for engines was realized before by one of the authors.

7

Experimental

1D Model

Figure 5: Comparison of Volumetric efficiency

Figure 6: Torque from various simulated manifold systems and DynoJet measured torque

8

3.3 Restrictor Modelling Steady state restrictor modeling on its own was performed and compared to data from Badih et al SAE paper 2001-01-2553. Flow bench data for our restrictor alone was not obtained during our flow bench tests as it was welded to our intake system. The model results showed that the diffuser section of the restrictor needed to be modeled as a tapered section in Wave so that enough mass flow rate is predicted. The nozzle section could either be tapered or square. Restrictor Mass flow rate 90.0 80.0

Mass flow gr/s

70.0 60.0

CorrectedPaperdata

50.0

Wavedata

40.0

2section Taper30mm TaperAutoCd

30.0 20.0 10.0 0.0 0.000

5.000

10.000

15.000

Pressure kPa

Figure 7: Restrictor mass flow comparisons

3.4 Other Model Considerations Cases We have our cases setup at 500 rpm intervals from 12000 to 3500 rpm with a number of specified variables. All modeling is for wide open throttle operation as this modeling effort is mostly intended to get better tuning of the engine for power considerations. So far we are not changing the variables in the different cases as we need experimental data to support our changes. Combustion Model So far we have been using the SI Wiebe function as our combustion model with the 50% burn point (THB50) fixed at 7o after TDC and 23o burn duration (BDUR) for all our Cases (i.e. different speeds). Other example models used these numbers even in high speed engine examples. Since we know our experimentally determined MBT spark 9

timing we would like to implement these in our model. The way to incorporate this seems to be to develop an IRIS model for the combustion. This task has not been undertaken yet as the model could be kept simpler if we obtain and use THB50 and BDUR that are relevant to high speed engines.

4

Closure

The Oakland university FSAE team is strengthening its powertrain knowledge by the combined use of experimental testing and simulation. Engine testing has been facilitated by the setting up of a dedicated dynamometer setup that is available for FSAE throughout the year. Donation of data acquisition system was solicited and it was setup with the dynamometer. Our own engine controller development was started and a prototype was tested with successful results, further testing is scheduled for the summer. 1D engine simulation was also started and the model will be used to design intake and exhaust system for tuning the FSAE engine. Cam timing will also be studied to decide if shifting the cams is beneficial.

Mario Farrugia Oakland University Formula SAE team May 11th 2004

10