22

BERT NEUHAUSCRISTOPH HINSE ET AL.

Many problems that can occur ininjection molding can be avoid-ed by simulating mold filling in

advance. Current commercial softwaretools are mainly designed to simulatethermoplastic materials. Simulating reac-tion-injection molding processes (RIM),though, is much more complicated,which is probably why there are as yet nosatisfactory solutions for this on the mar-ket. In view of this, BASF PolyurethanesGmbH, Lemförde, Germany, and Sim-paTec GmbH, Aachen, Germany, decid-ed to collaborate on developing a simu-lation tool for polyurethane (PU) RIMsystems.

The objective was to be able to predictthe filling behavior, including the fiberorientation, the shrinkage and thewarpage of the finished part, while allow-

ing for the chemical cross-linking reac-tion. The focus lay mainly on simulatingfast-curing,quasi-compact materials.Thedevelopment work was based on Mold-ex3D (manufacturer: CoreTech SystemCo., Ltd), a commercial software packagefor computational fluid dynamics.

Description of the SimulationMethod

The simulations are performed on a fieldpart and two other “test-plate molds.”Simulation comprises stages of fil-ling, hardening of the material and

RIM Simulation. So far, there has been a lack of simulation tools suited to

routine polyurethane RIM processes. This could be about to change: the tool

presented here has cleared the first test hurdles.

Simulating Reaction InjectionMolding Processes

The 75 % filling of aninjection molded frontspoiler correspondswell with the result ofa simulation (see bot-tom) (figures: SimpaTec,BASF Polyurethanes)

S IMUL AT I ON

© Carl Hanser Verlag, Munich Kunststoffe international 5/2010

Fig. 1. The Simulation of a 75 % filled front spoiler corresponds well with the real part (see top)

Translated from Kunststoffe 5/2010, pp. 46–48Article as PDF-File at www.kunststoffe-international.com; Document Number: PE110390

warpage. In order to be able to accurate-ly depict every aspect of what happensin reality, only 3-D simulations are per-formed.

The movement of the PU system,which becomes more viscous as timepasses, can be described mathematicallyin compliance with the laws of conserva-tion of mass, momentum and energy. Itis important in this regard to couple theinfluences of viscosity and the kinetics ofcrosslinking to each other so that a moreaccurate picture of the filling behaviormay be obtained.

Discretization of the laws of conserva-tion is effected by means of the finite vol-ume method (FVM), which is now usedsuccessfully in all areas of flow simula-tion. The flow front is calculated bymeans of the volume-of-fluid method(VOF).

Time-dependent Filling Study

For a detailed filling simulation, a knowl-edge of the following material parametersis needed:� The cross-linking reaction as a func-

tion of time and temperature,� the viscosity,� the thermal conductivity, and� the heat capacity,each as a function of the degree of cross-linking and temperature. Two different,established methods are used to charac-terize the kinetics of the RIM-PU system.First, a commercial IR spectrometer isused which has been converted for track-ing the polyurethane reaction underisothermal conditions. Second, the in-crease in temperature of the polyurethanesystem during the reaction is measured asa function of time,and the results are usedto determine the parameters of the reac-

tion kinetics under assumed adiabaticconditions. The method of adiabatic tem-perature increase [1] was used especially

for very fast systems. As part of the col-laboration, BASF Polyurethanes and Sim-paTec managed to develop a cross-linking

model and to implement this in the soft-ware.

The first experiment consists in simu-lating Elastolit R 8719/105LT, a poly-urethane which cures in about 2 s. Forthis, an automotive front spoiler filled to75 % is compared with the results of thecorresponding simulation. The real, in-completely filled part and the simulatedpart match in all important criteria (Titlephoto and Fig. 1).

To check whether the modeling de-scribes the cross-linking reaction in suf-ficient accuracy,“injecting into the cross-linking material” is performed, both inthe experiment and the simulation. Inother words, injection is performed soslowly that the material has cured beforethe mold has been completely filled. Thereal part can be removed after a fillingtime of 1.4 s. With the simulation, moldfilling stagnates after 1.37 s due to thecross-linking reaction and the resultantsharp increase in viscosity. The simula-tion is therefore very adept at reproduc-ing curing of the material during the shot(Figs. 2 and 3).

Fiber Orientation and Warpage

In the simulation, fiber orientation is de-scribed by the Folgar-Tucker equation,while the change in orientation to reflectthe flow conditions is described by themethod of Advani and Tucker.

To be able to validate the simulatedfiber orientation, experiments with a per-forated plate are carried out. The mate-rial used (grade: Elastolit R 4503) has afiber volume fraction of 15 %. Figure 4shows the fiber orientation in the fullyreacted perforated plate as measured byultrasound (across the plate thickness)[2]. The length of the arrows is indica-

Fig. 2. Partially filled front spoiler obtained by “injecting into cross-linking material” after1.4 s filling time

23

S IMUL AT I ON

>

Kunststoffe international 5/2010

Fig. 3. Simulation ofthe partly filled frontspoiler. After 1.37 sthe filling stagnates –a result that is almostidentical with the ex-periment

x

20

cm

16

14

12

10

8

6

4

2

00

y

2 4 6 8 10 12 14 16 18 20 22 24 26 28 cm 32

Fig. 4. Experimentally determined fiber orientation of a perforated plate with 15 % fiber by volume

© Kunststoffe

Flow direction

24

tive of the degree of orientation. Figure 5shows the simulated fiber orientation ofa perforated plate at the end of filling.Only areas of extensive orientation, anddegrees of orientation greater than 0.9(where 1 is optimum orientation), areshown. The comparison reveals goodagreement between simulation and ex-periment.

The warpage behavior of injectionmolded thermoplastics stems not onlyfrom the fiber orientation but also fromthe volume change of the matrix as afunction of pressure and temperature, i.e.pvT behavior. In reactive PU systems, thebehavior depends also on the degree ofcross-linking c. The pvT model musttherefore be expanded by the degree ofcross-linking to yield a pvTc model. Theassumption here is that the volume de-creases with increase in cross-linking.

Corresponding experimental meth-ods are being developed with a view tobeing able to describe the relevant pa-rameters of the cross-linking-dependentshrinkage regardless of thermal effects.The pVTc model is validated by mold-ing plate samples, both with and with-out fibers. The measurements are thencompared with the simulation results(Figs. 6 and 7). The simulation shows vol-umetric shrinkage of 0.45 % across thefiber and 0.25 % along the fiber. Thesevalues agree well with the measured val-ues of 0.48 % and 0.27 %.

Conclusion

As shown, the new model in the Mold-ex3D software package is a useful tool forsimulating fast-reacting PU-RIM sys-tems. Comparison between simulationand injection molding shows very goodagreement. The program can help usersto design components which are perfect-ly matched to the polyurethane system,and can be used for all compact poly-urethane systems. It also enables theprocessor to quickly and inexpensively re-spond to problems in tool design andparts filling.�

REFERENCES1 Macosko, C. W.: RIM: Fundamentals of Reaction

Injection Molding. Hanser Publishers, New York(1989), pp. 139–145

2 Predak, S. et al.: Faserorientierungsmessung ankurzfaserverstärkten PUR-RIM-Bauteilen: Kom-bination zerstörungsfreier Prüfmethoden zurOptimierung von Simulation und Herstellung-sprozess. Technisches Messen 73 (2006) 11,pp. 617–628

THE AUTHORSBERT NEUHAUS, born in 1967, has been project

manager in the Development Compact Systems de-partment (AD/C) at BASF Polyurethanes GmbH, Lem-förde, Germany, since 2005 and is responsible for PU-RIM development.

DR. MAX RÜLLMANN, born in 1972, was incharge of BASF’s Polymer Physics Unit in the GlobalPolyurethane Specialties Research Department atBASF Elastogran GmbH from 2005 to 2008 and hasworked in polymer research at BASF SE, Ludwigs-hafen, Germany, since 2008.

CRISTOPH HINSE, born in 1977, is CEO of SimpaTec GmbH, Aachen, Germany.

DR. REINHARD HAAG, born in 1960 is managingdirector of SimpaTec GmbH, Bangkok, Thailand.

BASF Polyurethanes GmbHD-49448 LemfördeGermanyTEL +49 5443 12-0www.pu.basf.de

SimpaTec Simulation & TechnologyConsulting GmbHD-52072 AachenGermanyTEL +49 241 9367-1500www.simpatec.com

Contacti

Fig. 5. Simulationof fiber orienta-

tion. The fiberalignment >91 %

is shown (bluemeans a high

degree of orien-tation)

Fig. 6. Simulation of part deformation in flow direction (z) Fig. 7. Simulation of part deformation across the flow direction (z)

S IMUL AT I ON

© Carl Hanser Verlag, Munich Kunststoffe international 5/2010

z z

z