|
|
 |
Misuse and Abuse of Plastic Process Simulation
Geoffrey Engelstein,GR Technical Services, Mountainside, NJ
Introduction
Mold filling, cooling, and warpage analyses are very powerful diagnostic and troubleshooting tools. However there are some limitations in what the simulations can be expected to do. Most bad analysis experiences arise from misapplying the technology to problems which are unsuitable. A wide variety of these limitations will be discussed, arising from simulation assumptions, algorithmic implementations, rheological data, and other sources.
Terms
There are three basic levels of plastic process simulation. In the most basic, only filling is modeled. An isotropic mold material is assumed, with balanced (and constant) temperatures in the core and cavity. Typically this is an appropriate assumption, and allows for evaluation of injection conditions, such as weld lines location, stress levels, and pressure requirements.
For the second level, or cooling analyses, a more detailed model of the mold is added, which can include cooling channels, baffles, bubblers, heat pins, inserts, parting lines, and other mold features. Here the analysis evaluates the mold temperature rather than treating it as constant. While this allows for a more detailed look at the cooling cycle, it also improves the accuracy of the filling stage, as mold temperature variations can improve or impede material flow.
Warp analysis is the final level, incorporating results from both the filling analysis (mainly density variations) and cooling analysis (temperature variations) to determine the forces that act on the part after ejection.
Two-shot Molding and Inserts
Two-shot molding is becoming a more widely used process. This leads to interesting challenges and pitfalls for simulation. While the first shot can be modeled as a normal part, the second part has to take into consideration the prior plastic molding. In this situation the 'part' which is modeled occupies the space of the second shot only. The first molding is simply treated as part of the mold steel. If only a filling analysis is performed on this model inaccurate results are often the result. Plastic, of course, is an excellent insulator, so while the mold steel is transporting heat out of the cavity, the first shot is keeping it in. This goes against the base assumption of isotropic mold behavior.
In these cases it is imperative to use a more sophisticated level two analysis, and model the first shot as an insert with a low thermal conductivity. However, even this can sometimes break down, especially if the first shot has a melt temperature comparable with the second shot. This softening of the first shot can alter its thermal transport properties and reduce accuracy.
Since warp simulations are based on the effects of temperature variation in the cavity only, two shot warp analyses are virtually impossible.
Metal inserts present many of the same difficulties, however filling analyses tend to be more accurate as the thermal conductivity of the insert is typically comparable to that of the mold steel. However, there are some situations which demand a more sophisticated analysis. If the insert is encapsulated (or nearly encapsulated) it can act as a heat sink which will allow improved flow. This usually leads to pessimistic results from the simulation, such as predicting higher injection pressures and clamping required than will actually occur. Often by examining the part geometry it will be obvious that this effect will be minimal, if, for example, the insert is at the end of a flow path. If accuracy is needed it is necessary to model the inserts explicitly in a full cooling analysis.
For the same reasons as in two-shot moldings, warp analyses are basically impossible to perform on parts with inserts, unless they are small or are far from the area of interest.
Marbling and Mixed Materials
Every other month or so we will get an inquiry from someone who is molding marbled pieces. The granite look is also very popular these days. Typically they are getting poor swirling, and the parts are coming out with the colors clumping and badly dispersed. Plastic process simulation is currently unable to model these types of situations, and cannot be used to correct these problems. In fact, the turbulent processes that create these effects are explicitly not part of the simulation, as will be discussed in the next section.
Bulky Parts (Thick walls)
The basis of filling analysis is the assumption that the plastic flows in a sheet-like manner, with the wall thickness being small compared to the overall dimensions of the part. This is reflected in the finite-element models of the parts, which are made up of two-dimensional elements. As parts and features become more 'cube-like' the modeling problems become more severe. Turbulence, eddies, and swirling backflow can all occur in extremely thick parts, but none of these behaviors are included in the simulation. The results can be very inaccurate, even basic flow results, and can lead to poor decisions about part and mold design.
Linking Process Simulation and Structural Simulation
Plastic process simulation can pinpoint weak areas in parts caused by processing, such as weld lines and areas of high molded in stresses. Often it is desirable to see how these defects will effect the eventual strength of the part under various loading conditions. Unfortunately, there is no mechanism for transferring filling and cooling results into a structural simulation. Structural analysis typically assumes isotropic material behavior, an assumption which is rarely accurate for molded parts. Part defects can be modeled manually, but the numerical estimation of, for example, how a weld affects part strength is problematic at best. If attempted, this leads to false confidence in the structural results.
A better solution is to perform the structural analysis with an isotropic material specification, and ensure that any molding related defects are located in low stress areas.
Effects of Additives
Additives can dramatically impact a resin's behavior. Often they are purchased from a third party and blended in at the molding facility. Unfortunately analyses are typically performed on the basis of data supplied by the resin supplier, which usually does not include additives. All molders have experienced the differences that can result just from molding parts in different colors. Yet very rarely is rheological data available for these blends. So the simulation begins on shaky ground.
Luckily color additives usually do not effect flow or thermal characteristics. Thus the basic filling and cooling analyses retain their accuracy. However shrinkages can be greatly impacted, making warp estimates prone to error. If warp and dimensional stability is important it can be worthwhile to have testing done on the colorized material to determine its properties.
Unlike colorants, other fillers like carbon fiber, glass, flow enhancers, and flame extinguishers, can have a dramatic impact on flow. Fortunately, these are usually added by the resin supplier and so material data is readily available.
Evaluating Aesthetics
Often one of the primary purposes of a process simulation is to evaluate the final aesthetics of the part. How visible will the welds be? Will I notice the gate blush? Will the sink marks be bad? To a degree all of these questions can be answered by analysis, however the answers tend to be 'pretty bad' or 'pretty good', which can make a go/no-go decision on a project difficult. While there are direct numerical results that can be used to evaluate aesthetic problems (like weld incidence angles and shear stress), the final 'visibility' of a defect is greatly impacted by part color and texture.
Rather than try to determine exactly what a defect is going to look like, a better plan is to try to minimize the problem or shift it to a different location on the part.
Warp and Dimensional Stability
As the most complex analysis, warp is also the most subject to error. Our experience has been that warp simulation is most useful on a qualitative, rather than quantitative basis. The results can be used to determine what the warp tendencies of a part are and what is causing them, but the numeric warp result is suspect. There are a variety of reasons for this.
Material data is one, as it can be difficult to get all of the information needed for an accurate warp response. Transverse and linear shrinkages, non-linear stress-strain response, and other data are not usually measured by resin suppliers, and outside testing can be expensive.
Another problem can be caused by variation on the shop floor. Often small temperature variations (on the order of 20EF) between the core and cavity are the entire basis for the bending moment. It is not unreasonable to expect this difference to vary five degrees in the press, whether from miscalibrated chillers, high ambient air temperature, or many other causes. This results in a 25% difference in the predicted difference, which will skew the actual measured warp.
Also, the simulation bases are partially at fault. Warp results from a complex interaction of a myriad of factors. The interrelationships between these factors and basic physics is just not that well understood at the present time. This will improve in the future. However, warp simulation is still valuable, both for the qualitative information and for the ability to compare various scenarios to determine which is best.
Mid-Plane Modeling
As stated earlier, plastic process simulation is based on two dimensional finite elements. Each element has a wall thickness associated with it. This is obviously an approximation of the actual part, and significant error can be introduced by modeling mistakes. This is a detailed area which has been well covered in previous papers (Engelstein Antec92 and G. Batch Antec93), so only a few key points will be covered here.
First, weld lines will only be seen on element boundaries. So it is important to select element size with this in mind. Second, gate modeling techniques are very important. Mismodeling a gate, either by using elements that are too large or too small can result in poor results, especially regarding pressure and shear. Similar care needs to be taken around living hinges and other significant wall thickness transitions.
Material Data Quality
The need for good material data cannot be overstated. As analyses become more complex, more parameters are needed to accurately describe resin behavior. The best place to get this data is from the resin suppliers. However, often their data is not of the best quality, whether because of testing inaccuracies, limited data sets, or misapplied correction factors. It is best to try new data with a sample part you are familiar with. This will allow you to see if the results look close to those obtained with a similar material.
Conclusion
Despite all of the negatives discussed above, plastic process simulation is an extremely valuable tool for the plastics engineer. It gives insight into the molding process, allows for optimization of part and mold designs, and helps to get it right the first time. It is important to understand the basis and limitations of these analyses so that it can be used in appropriate cases with appropriate expectations.
|
|
|
