2006_001_01

8/13/2019 2006_001_01

1/11

24

th CADFEM Users Meeting 2006

International Congress on FEM Technologywith 2006 German ANSYS Conference

October 25-27, 2006 Schwabenlandhalle Stuttgart/Fellbach, Germany

1

Virtual Development of Four-cyl inder Diesel Engine

Vaclav Pistek, Pavel Novotny

Brno University of Technology, Institute of Transport Engineering,Brno, Czech Republic

Summary

Modern powertrains are required to have high reliability and power, low consumption, emissions andvibration level. The internal combustion engine development process requires CAE models whichdeliver results for the concept phase at the very early stage. The vibratory and acoustic behaviourof the cranktrain is a highly complex one that greatly effects global engine noise emissions. The paper

deals with the cranktrain virtual development of a new series of a tractor in-line four-cylinder dieselengine. Experimental verification of the computational model results has been performed bymeasuring the values on a testing engine unit using non-contact laser measuring technique.

Keywords

diesel engine, cranktrain, crankshaft, vibration, fatigue

8/13/2019 2006_001_01

2/11

24




2

1. Introduction

Manufacturers in the military automotive industry are constantly looking for ways to maintain theircompetitiveness by getting innovative products to market first. Plus, safety issues, fuel efficiency,warranty costs, styling and other goals place increased demands and expectations on the militaryautomotive industry to develop high-quality products. Pressures to meet these stringent requirementswhile reducing product development cycle time on the order of 50 percent represent a huge challenge

to researches, suppliers, co-development partners and others in the military automotive value chain.Time-to-market is critical, because the value of innovation will be lost if competitors win the race to theshowroom floor with similar designs. A few weeks delay can result in millions of dollars in lostrevenue, and the first company reaching the market typically gets million share of sales. As a result,more and more manufacturers are discovering that the only way to achieve development goals is toemploy engineering simulation technologies. This goal can only be achieved by increasing the amountof simulation. All major military truck producers, as well as companies in a wide range of othermanufacturing industries, are using virtual prototyping to evaluate designs through computersimulation and analysis, thereby reducing their reliance on hardware testing. The aim is not to entirelyeliminate such testing, but rather to reduce the number of cycles and guide the empirical proceduretowards verifying well-defined performance attributes. By studying product configurations in the earlyphases of development, engineers can make changes and refine designs easily and inexpensively.Studies have shown that the cost of change increases exponentially with each stage of development,

thus making changes costly during detailed design, very expensive during prototype testing andtremendously high during production. Moreover, correcting errors after the product is sold can beprohibitive in terms of recall and warranty costs, sometimes causing economically catastrophicconsequences for manufacturers. One of the greatest benefits of simulation early in development isthat engineers have the time and technology to evaluate alternatives, run what-if scenarios andcome up with innovative designs. In a major push to accelerate its product development cycle, onemajor automaker implemented a concerted front-end loading initiative involving the use of earlysimulation in conjunction with other process changes and slashed development time and costincluding the number of full physical prototypes by between 30 and 40 percent.

2. CAE Modeling

2.1 Connecting Rod Design Modifi cation

The new series of a tractor in-line four-cylinder diesel engine will be using a modified versionof a connecting rod. The mass of the modified connecting rod is more a 28 percent less,the reciprocating weight is about 46 percent less and a rotating weight is 7 percent less as well.

Fig. 1: Design modification of connecting rod: a) original version b) modified version

8/13/2019 2006_001_01

3/11

24




3

Figure 1 shows an original version and a modified version of the connecting rod. Present cranktrainintegrates a balance unit but this unit balances only 86 percent of second order unbalanced forces. Adecrease of the connecting rod reciprocating masses causes a full balancing of second orderunbalanced forces. All main construction sizes, for instance an eye spacing, screws etc. hold theoriginal connecting rod line.

2.2 Implementation of CAD and FEM into MBS

For complex simulation processes and high-quality products such as combustion engines, highereffort is necessarily required for simulation and modeling. The development process starts withComputer Aid Design (CAD). For the cranktrain dynamics CAD models give basic geometricinformation about engine parts and can be used for the creation of FE models (e.g. crankshafts orengine blocks) or MBS rigid parts (e.g. pistons or piston pins) [2]. Figure 2 shows a CAD model of anin-line six-cylinder diesel engine cranktrain including all significant moving parts.In the powertrain development process, several numeric simulation techniques have steadily gained inimportance. In the case of dynamic structural calculations the Finite Element Method (FEM) is state ofthe art.In many cases the FE models are suitable for dynamic calculation such as modal analysis or harmonicanalysis, but the durability of the product requires a different FE modeling approach. Frequentlydynamic models have also been simplified to remove details such as holes and fillets which play animportant role in the durability of the product. FE model of in-line six-cylinder crankshaft includes small

geometric details and is suitable for a stress-strain analysis as well as a fatigue analysis. For realisticcalculations the model of a FE block must be attached but the FE model of an engine block is used formodal analysis and for stiffness calculations. It therefore has a uniform hexa element mesh. Figure 2presents a detailed FE models of an in-line four-cylinder crankshaft and a connecting rod.

Fig. 2: CAD model of four-cylinder diesel powertrain moving parts

8/13/2019 2006_001_01

4/11

24




4

Crankcase stiffness also significantly influences cranktrain dynamics. As the stiffness of the crankcaseis higher, the crankshaft vibrations are lower. Mainly bending vibrations of the crankshaft are verysensitive on crankcase stiffness. Torsional vibrations of the crankshaft are partially influenced as well.If the stiffness is decreased then the torsional natural frequency rises up and vibrations decrease. Thecrankcase stiffness is included into MBS model as well.

Fig. 3: FE models of in-line 6-cylinder diesel engine crankshaft and connecting rod

Complex computational models of powertrain are assembled in Multi-Body System (MBS) ADAMS.MBS enables the creation of virtual prototypes, realistically simulating the full-motion behavior ofcomplex mechanical systems on computers and quickly analyzing multiple design variations until anoptimal design is achieved. This reduces the number of costly physical prototypes, improves designquality, and dramatically shortens product development time.One drawback of the MBS program is that all components are assumed to be rigid [5]. In the ADAMS

program, tools to model component flexibility exist for geometrically simple structures only. To accountfor the flexibility of a geometrically complex component, ADAMS relies on data transferred from finite-element programs such as ANSYS.The flexibility of a geometrically complex component, such as a crankshaft, is represented by a matrixthat is stored in modal neutral file (.MNF) created by a FE program.The algorithm used to create the modal neutral file is based on a formulation called component modesynthesis (also known as dynamic substructuring). ADAMS uses the approach of Craig Bampton withsome slight modifications. According to this theory, the motion of a flexible component with interfacepoints is spanned by the interface constraint modes and the interface normal modes. Constraintmodes and interface normal modes together are referred to as component modes.Since the algorithm relies on the component mode synthesis method, which is based on the modalanalysis, only linear properties are considered during the formation of the modal neutral file. Allgeometric and physical nonlinearities are ignored. If significant geometric nonlinear effects are presentin your component, you must subdivide the component into several smaller components and transfereach one separately. You can then assemble the subdivided components in MBS to form a flexiblecomponent with geometric nonlinearityThe parameters which influence the accuracy of the reduction are the number and location of theboundary nodes and the number of first modes considered during the reduction. These parameterscan be validated, for example, by computing a modal analysis of the FE model with constrains inANSYS and separately in ADAMS and by subsequent comparison of the results.

2.3 Slide Bearing Model

A slide bearing is described as a sleeve around a pin with a lubricating fluid. The lubricant is suppliedwith a suitable slot. Tangential and radial motion, in combination with a wedged gap, generate a

8/13/2019 2006_001_01

5/11

24




5

pressure in the oil film in the slide bearing. Bearing loading is periodical and a pin center comesthrough a bearing trajectory.The dynamic behavior of hydrodynamic bearings is described by the Reynolds differential equation.The Reynolds differential equation cannot be solved analytically. The three-dimensional method isused for the slide bearing model that includes misalignment effects. The Reynolds equation must besolved explicitly. To keep the simulation effort in a reasonable range, the hydrodynamic solution mustbe decoupled from the dynamic solution of MBS Solver. Therefore, the Reynolds equation is solvedfor several bearing working conditions (eccentricities and misalignment angles) prior to the dynamicanalysis. The results are stored in hydrodynamic databases representing dimensionless bearingreactions (forces and force-attachment coordinates) to dimensionless states (eccentricity andmisalignment value). During the dynamic solution, MBS Solver subroutines do the database accessand the necessary analytical steps (coordinate transformations, and so on) [4].

3. Measurement Technique

The comparison of calculations and measurements is still a decisive quality criterion in the CAEenvironment. Non-contact measurements with laser Doppler vibration meters enable verification ofcomplex calculation models. The torsional vibration meter B&K Type 2523 (see Figure 4) wasdesigned for making torsional vibration measurements in places where mounting a transducer onto arotating object is not possible.

Fig. 4: Laser torsional vibration meter B&K Type 2523 (left) and Laser vibration meter B&K Type 3544(right)

The laser torsional vibration meter, its principles and applications are described below. A schematicdiagram of the electronic and optical components in the torsional vibration meter is shown in Figure 5.The central module of the system is a low power Ga-Al-As laser. The laser beam is split into twoequal-intensity parallel beams separated by a distance

BAAA RRd coscos +=

. (1)

The beams strike the shaft surface at points A and B, where the velocity is VA and VB, respectively.Each beam sees only the velocity in the x-direction

xAAxAAA VRVVv == coscos (2)

xBBxBBB VRVVv == coscos (3)and is thus frequency-shifted

AA

vf

2=

and

BB

vf

2=

(4)

8/13/2019 2006_001_01

6/11

24




6

Fig. 5:The arrangement of the electronic and optical components within the B&K Type 2523

The returning beams then heterodyne, giving an output current modulated at the beat frequency, fD,that is, the difference between the frequencies of the Doppler-shifted beams

dvvfff ABABD

2)(

2===

. (5)

Thus, the beat frequency is directly proportional to the shaft speed and is independent of any solidbody motion (VX + VY) of the shaft. If the plane of the laser beams is not perpendicular to the shaft

axis, then fDis also a function of cos , where is the angle between the plane of the laser beamsand the plane perpendicular to the axis of shaft rotation.

Fig. 6: Filter characteristics

The torsional vibrations of the six-cylinder in-line diesel engine have been measured and evaluated byin-house software. The measured signal has been processed by the averaging filter

8/13/2019 2006_001_01

7/11

24




7

=

=1

0

)(1

)(K

k

kTtxK

ty

(6)

whose characteristic

=

=

+

0

0

00sin

sin

1

f

ff

fK

KffGi

ffG

, i = 1,2, (7)is periodical. For filter characteristics see Figure 6.

4. Vibration of cranktrain Results

Failures and mainly fatigue fractures occurred with an engine power increase and material savings inthe past. These fractures were not caused only by forces from combustion pressure processes andinertial forces or incorrect construction but it was found that the main portion of these fractures wascaused by periodical torsional vibrations of the crankshaft. Three types of crankshaft vibrations exist in

agreement with traditional approach: bending, torsional and axial vibrations. Bending vibrations arecaused by periodical forces that operate perpendicular on the crankshaft axis. These forces are radialand tangential forces on the crankshaft and imbalanced eccentric forces of the cranktrain. Naturalfrequencies of bending modes are given a free length of the crankshaft between the bearings. Freelengths of crankshaft between two bearings are very small and natural frequencies of bending modesare high and there are no resonances at an engine speed range. Stiffness of the bearings and theengine block has a large influence on bending natural frequencies.Torsional vibrations of crankshafts are much more dangerous than bending vibrations. Forcedtorsional vibrations of the crankshaft are caused by time depended torques. Torsional vibrations reachhigh values in resonances when the frequency of forced vibration is equal to natural frequency. Theresonances and relevant critical engine speeds cause an increased noise and engine vibrations. Allvibration types can be obtained from a complex computational model of the cranktrain.The complex computational model of the in-line four-cylinder diesel engine was developed and verified

by measurement on an experimental engine with attached Schenck dynamometer type W 400.Results of vibration analysis are shown in figures 7 and 8. Figure 7 presents peak-to-peak values oftorsional vibrations of the in-line four-cylinder diesel engine cranktrain without a damper vs. enginespeed and Figure 8 presents the torsional vibration order analysis of the in-line four-cylinder dieselengine cranktrain without a damper vs. engine speed.

Fig. 7: Torsional vibration of in-line four-cylinder diesel engine cranktrain peak-to-peak values

8/13/2019 2006_001_01

8/11

24




8

Fig. 8: Torsional vibration order analysis of in-line four-cylinder diesel engine cranktrain

Vibration analysis results provide input data for the fatigue analysis of a crankshaft. The results showthat engine run is possible without a torsional damper. A presence of some type of a torsional damper(viscous of rubber one) would decrease vibrations and noise produced by the cranktrain and will be

considered for the new series of the tractor in-line four-cylinder diesel engine.

5. Fatigue Analys is

As a part of the engineering design process, engineers have to assess not only how the designsatisfies the performance requirements but also how durable the product will be over its life cycle. Amajor cause of failure is the growth of cracks that grow due to fatigue loadings to the point where theproduct fails. This part describes an approach to predicting fatigue life and combines FEM and MBSsystems.Engine performance has to be kept for the whole engine service and therefore the decisive criterionfor the most stressed parts is the fatigue analysis.Fatigue calculations of a dynamically loaded crankshaft of an in-line four-cylinder diesel engine arecarried out on the basis of finite element calculation results. The starting point for the multi axial-fatigue analysis is a time series of stress tensors for a specific node. Critical cutting plane methodsand material sensitive equivalent stress hypotheses are used to count rainflow matrices. Damage iscomputed from the rainflow matrices taking many fatigue influences into account (notches, meanstresses, temperatures etc.)High cycle fatigue can be expressed by using endurance safety factors. The endurance limit safetyfactor can be defined as

ma

MASF

+

+= , (8)

where A allowable amplitude,M mean stress,a actual occurring amplitude.

8/13/2019 2006_001_01

9/11

24




9

A minimum endurance limit safety factor (MSF) is the minimum value of all endurance limit safetyfactors of the FEM model and can be expressed as

{ }iSFMSF min= . (9)

A minimum endurance limit safety factor of e.g. MSF = 1.2 with regard to the upper stress isinterpreted so that one could increase the total stress (amplitude + mean stress + possible constantstress) by a factor of 1.2 in order to achieve an endurance limit safety factor of MSF = 1.0, all otherthings being equal.

5.1 Crankshaft

Crankshaft fatigue analyses are performed for several steady state solution results with an enginespeed step of 50 rpm. A crankshaft surface distribution of an endurance safety factors in Figure 9 ispresented for a crankshaft without a torsional vibration damper and for the engine speed of 2200 rpm.

Fig. 9: Crankshaft fatigue analysis results SF surface distribution for engine speed of 2200 rpm andSF on radiuses vs. engine speed

8/13/2019 2006_001_01

10/11

24




10

Crankshaft fatigue results are in acceptable level and no crankshaft design modifications have to beperformed.

5.2 Connecting Rod

There are several loading sources: combustion pressure, inertial forces and pretension screw forces.Inertial forces change with square of angular velocity and combustion pressure forces change only

slightly with engine speed. Pretension forces can be treated as an engine speed independent. Takinginto account that there is no connecting rod resonance in the whole engine speed range, only onesteady state nonlinear solution is performed. The engine speed of 2200 rpm is chosen for thecalculation. The results of FEM analysis are used for fatigue calculations.Figure 10 presents results of a fatigue analysis of an original (Figure 10a) and a modified version(Figure 10b) of a connecting rod.

Fig. 10: Connecting rod fatigue analysis results SF surface distribution for engine speed 2200 rpma) original version b) modified version

Generally a fatigue of the modified connection rod slightly decreases but is still in safety limits.

6. Conclusion

Virtual prototypes of engine parts make it enable study and refine real-world product performance.These are often multiphysics analyses for studying the interaction of all factors encounteredthroughout the overall product development, including real-world factors that radically influenceperformance such as temperature, vibration or and fatigue. Such simulation may determine thedeformation of the crankshaft and connecting rods, for example. This approach overcomes the historicbuild-test-redesign problems by evaluating designs through computer simulation and analysis earlierin the product development process and reducing reliance on validation testing late in the cycle. Oftenperformance problems that are encountered late in the product development cycle necessitate

8/13/2019 2006_001_01

11/11

24




11

repetitive redesign cycles until satisfactory performance is achieved, with several testing iterationsusually required. This adds considerable time and cost to the development cycle.

Conclusions

Published results were acquired using the subsidization of the Ministry of Educationof Czech Republic, project 1M6840770002 Josef Boek Research Centre for Engine and Vehicle

Technology II.

References

[1] Novotn, P.: Cranktrain dynamics simulation central module of virtual engine, Ph.D.Thesis,Brno University of Technology, 2005, ISSN 1213-4198

[2] Pistek, V., Novotny, P.: Virtual Engine - A Tool For Reliability Increase, Reliability 2004,University of Defence in Brno, October 2003, ISBN 80-85960-77-X

[3] Rebbert, M.: Simulation der Kurbewellendynamik unter Bercksichtigung derhydrodynamischen Lagerung zur Lsung motorakusticher Fragen, Ph.D. Thesis, Fakultt frMaschinenwesen der Rheinisch-Westflischen Technischen Hochschule Aachen, 2000

[4] Shabana, A.: Dynamic of Multibody System Second Edition, Cambridge University Press,Cambridge, United Kingdom, 1998, ISBN 0 521 594464

2006_001_01

Documents