A step-by-step CAD exercise in Solidworks on an exercise to build a plastic game console. Highlighting drafts, fillets, cuts, patterns (linear and circular) as well as using construction/ other geometry to highlight DESIGN INTENT… this is an example of my CAD ability (I also know Autodesk Inventor). I’m on a job search north of Boston, MA; what could I do for you?
I’m going to discuss two situations from my career where an obsession with the price per pound of the resin was evident, rather than the much-more-relevant price per part. I’ll conclude with some points to ponder when choosing plastics for cost minimization, whether in a new application or when considering substituting one resin for another in an existing product.
The first situation was my job straight out of graduate school. I’d landed at the engineering center for M.A. Hanna Resin Distribution, now PolyOne. I’d come to this particular project quite late, but still managed to contribute a little, and to see the end-game play out.
The situation was that a nationally-known maker of home laundry-drying racks, er, exercise equipment 🙂 who wanted a plastic base platform underneath the treadmill of their latest design. Their purchasing manager had heard of a material we had available: recycled polypropylene battery trays. It was cheaper than dirt – if I recall, something like 20 cents a pound. It also had the engineering properties of dirt. Low stiffness, low strength, low modulus of elasticity, and being recycled with varying feedstock, it processed inconsistently with predictable quality and consistency issues. But it was cheap.
So despite our concerns about this material’s suitability for the application, we had developed a design that met their specifications. It was heavy, with many deep ribs and thick sections necessary to meet the structural, deflection, and impact requirements. Because of these design features, we estimated its cost per part as fairly high despite the low cost of the base material.
As an experiment, we picked a prime engineering resin; I recall it being a PC/PET blend with 10% glass fiber reinforcement. At close to 10x the per-pound material cost of the original material, it seemed like a non-starter. But we plugged the different loading scenarios into the CAD program’s design optimizer (at this employer I used Pro-Engineer), with parameters such as the number, depth, and thickness of ribs, the thickness of the base platform’s flat area, and so on. Setting the objective function to minimize the part volume while still meeting all the different loading scenarios, we let slip the dogs of optimization to try different design iterations.
Lo and behold, because we were using a prime resin with much higher strength, rigidity, and impact resistance, the reduction in the thicknesses, depths, and number of ribs resulted in a part that was so much lighter than the original design that the reduction in weight more than compensated for the higher per-pound price. Additionally, there were three other cost benefits; one we could approximate, and the other two were an arm-waving savings we couldn’t quantify, but which were definitely something that needed to be considered.
First, because the mass of material was so much lower, and the wall thicknesses so dramatically thinner, the estimated cycle time per part was vastly reduced because the part could be cooled more quickly; this meant – going from memory – something like a 30% increase in the number of parts per hour. Even if the material cost had been a wash from one to the other, this added in a second advantage for the better material.
Second, the because of the weight reduction, shipping costs would be lower because of the reduced weight and lower volumetric part envelope, which resulted in a greater packing density per shipping container. We couldn’t quantify this in any meaningful way, but it was certainly something to point out. And third, the lower number of ribs, thinner sections, and overall shallower design meant less machining of tool steel, for an unknown but definite advantage in tooling cost and timing coming from the more expensive material.
But the project was killed. In our presentation to our customer we had two columns for the two materials. And we made the mistake of having, right under the two material names, the price per pound. Our customer’s purchasing manager never got past those two numbers to the nitty-gritty where the part with a more expensive resin was actually less costly, with other benefits to boot.
The second example comes from when I was at Ford Motor Company in Sandusky, Ohio; specifically, injection molded nylon housings for air cleaners. There were two suppliers (“A” and “B”) who continually vied for the business – we bought millions of pounds of plastic a year just for this application family. Big bucks were at stake.
Our purchasing person was obsessed with price-per-pound. Company A had had a cent-or-two advantage, and this was the supplier they wanted to use. But Company B had three advantages, and I (and others!) wanted to use them preferentially.
First, the densities were different. Company B’s material was lighter; even though it was marginally more expensive per pound, since the mold’s cavity had the same volume of plastic used the relevant parameter was not cost-per-pound but cost-per-cubic-inch. Company B’s material, on that basis, was actually roughly a cent per part cheaper.
Company B’s material also processed marginally faster. Because it was slightly less dense, and had a fractionally-higher thermal conductivity, it would cool faster after injection, leading to a couple of extra parts per hour.
And they had one final advantage: Company B was much more responsive. Faced with a service request call because of a production issue, Company A’s response was, typically, to schedule a visit within a week or so. Company B? You’d call with a problem and they’d be there the next day – live and in person – to help you figure things out.
Presenting these arguments convinced Purchasing to go with Company B.
From these two examples I’d like to abstract out some lessons which I hope are useful:
- In the design phase of a project, use your CAD program’s optimization feature to minimize the volume of material used under the different loading scenarios. It’s very likely that the cheapest material, which probably has lesser properties than a more expensive one, may require more material than the expensive one; the reduction in material used per part might well overcome the more expensive material’s price per pound sticker shock.
- Cycle time is important too. A higher-cost material, having better structural properties, can reduce cycle times by having thinner walls, which cool faster; this adds more parts per hour into the cost equation.
- Depending on the part’s functionality, a higher-end resin may require less structural features like ribs and gussets, as well as those features being smaller – resulting in a less expensive tool delivered faster because of reduced machining requirements.
- When considering swapping out one material for another in an existing mold, consider these two “lumped parameters” as rough first-pass screening tools; these two will interact, and your internal labor cost will be necessary in factoring out which one is more important (remember – these are presented as screening tools only, not definitive factors – you need to do a proper analysis based on your own situation!):
- Multiple price per pound times density to get price per volume. Using this parameter, the lower cost part will come from the material with the lower value.
- Multiply density times heat capacity and divide by the thermal conductivity. Using this parameter, the lower cost part will come from the material with the lower value.
- When presenting alternative materials, especially to non-engineers, put the estimated price per part right at the top of the two columns comparing the alternatives – above any other data. Get into the details of price per pound, wall thickness, cycle times, etc., later to support your cost per part estimate. Remember that, at the end of the day, what matters is cost per part. So put that first!
- Recycled materials are not necessarily bad materials per se, and can often offer substantial cost advantages – but unless the supplier takes extreme care in the recycling and pelletizing operation they may have the requisite physical properties while processing less consistently from lot to lot. This inconsistency may end up being more trouble than the lower material cost is worth. (This was seen in a third case at Ford, not discussed above, where we considered recycled polycarbonate from CDs instead of virgin material; the inconsistent processing and increased scrap eliminated any material cost advantage.)
- Service and response time matter. When you have a problem you need help now, not in a week or so. Lost production can cost you a lot, in scrap costs and in OT required to make up lost production, as well as in your reputation with your customer (and possible penalties they may charge you for not meeting their schedule). That responsiveness is worth an added margin to the raw material cost – especially in these days of lean manufacturing and minimal safety stock which could otherwise insulate your customer from your production hiccups.
Update: Thanks to Deepak Ramanathan, who pointed out the potential that a lighter part – depending on other functional requirements, of course – might also be moldable on a smaller machine, thus reducing costs in another way!
© 2014, David Hunt, PE
While working at Ford Motor Company in Sandusky, Ohio – aside: home of the incredible amusement park Cedar Point – I worked with two other engineers on “Integrated Molding and Assembly” (IMA) as a new technology to replace using glue in automotive lighting. Our first application was on the Ford Econoline headlight, as highlighted here on a portfolio page (alongside a small test piece, which I reference below). This process was patented; my first patent!
Essentially, this process molded a seal bead of plastic, in the case of the Econoline made from polycarbonate, to join two other pieces – the lens and the body, also polycarbonate – and form a bond that was strong and leak-tight. The process also worked with using acrylic and ABS bodies and either acrylic or ABS as the seal bead. In almost all cases, burst testing cracked the substrate parts before the seal joint broke. Descriptively, this was an overmolding process – a seal was overmolded onto the two substrate pieces.
Giddy with success including a successful launch on a real product (the Econoline), and as the plant was casting around for other potential product lines to bring in to diversify the plant’s product mix, we took a long look at nylon intake manifolds. The idea was that two halves of the manifold would be molded, and in either a rotary or shuttle mold the two parts would then be joined, and a seal bead overmolded; resin pellets in, complete manifold out. The patent space was already crowded for this product and similar process technologies, but we and the Ford attorneys believed we had something unique that could be done in a patentable way.
So we molded a bunch of lenses and bodies in our little test mold (again, see my portfolio page, link is above) made out of nylon 6/6 to test the concept; this being the standard material for the application. We put them in the mold, overmolded a seal bead out of the same material… and we could pop the pieces apart with our hand. There literally was no adhesion between the seal bead and either of the base parts.
Panic set in. Everyone was wondering what was going on; it had worked so well before. And then… I had an insight. All of the materials we had molded and sealed with enormous success were amorphous plastics. Nylon, in contrast, is a crystalline plastic. There is a huge difference between these in several dimensions, and the one that mattered was heat capacity; crystalline materials typically have a higher heat capacity because of the material’s microstructure.
For a strong bond in an overmolding application, the overmolding material needs to remelt a microlayer of the substrate to which it can then bond. Because crystalline materials have a higher heat capacity, more energy was needed to form that remelted microlayer when we had nylon parts; more energy than the molten seal bead of the same material had available.
To test this theory I arranged to use Amodel, a material that’s similar to nylons chemically but with a higher processing temperature, as the seal bead on our nylon 6/6 substrates. This time it worked; the assembly didn’t fall apart. Clearly I was on the right track; it was the crystalline nature of the substrates that was the issue. When we switched to nylon 6 for the substrates, a nylon also used – though not as commonly – for intake manifolds, it worked even better… because nylon 6 has a lower melt temperature and heat capacity. This allowed for much better formation of that necessary remelted microlayer.
The project was now a “GO!”, but we found out that in the course of talking with one supplier, we had not obtained a non-disclosure agreement (NDA) first. Ford’s lawyers said this ruined any patent possibility we might have had. Without the ability to legally protect the technology, the project died.
So here are the key points to take away from this essay:
- When overmolding, amorphous materials are generally easier to overmold onto than crystalline ones.
- If overmolding onto a crystalline material, make sure the overmolding material is molded hot enough to overcome the higher heat capacity inherent in the substrate’s crystalline structure (even if amorphous, mold the lower melt temperature material first, and use the higher temperature material as the overmolded material).
- When overmolding, the more similar the overmold and substrate materials are chemically, the better; here’s a good Rule of Thumb: If they can be alloyed (e.g., PC/ABS alloys), they’re similar enough for a good overmold application.
- I strongly recommend adding features to physically “back up” the ability of the two materials to engage with each other from overmolding and thus resist whatever forces will be applied. I.e., use keyway features, flanges, and grooves to form a physical interlock.
- When you have a concept that you even think might be patentable, make sure you protect yourself with NDAs before you talk with anyone outside. A few days of delay while getting things signed can save a new technology from being ruled not patentable.
© 2014, David Hunt, PE
When I was in graduate school at Carnegie Mellon University I did my master’s research in Design for Assembly (DFA) using Fitt’s Law as the centerpiece, and Boothroyd Dewhurst as a supplemental methodology. One of the things I researched was the idea of partial symmetry; essentially, there are often things are almost symmetrical, but not completely symmetrical. Why is this important? Because when an object is truly symmetric, it can be assembled in any number of ways (e.g., a square peg into a square hole because there are four orientations that work vs. a peg with a duck profile into a duck profile hole).
Often, however, things can be almost symmetrical visually but not quite enough to be assembled in multiple ways. When assembling such a component, that asymmetry must be detected so that the assembly can be done correctly; sometimes that asymmetry can be very difficult to detect. And in the worst case, there might a situation where something could be assembled multiple ways, but only function in one. My research on partial symmetry – part of my overall research – resulted in a paper:
Hunt, D.O. & Sturges, R.H., 1994.“Detection and Evaluation of Planes of Partial Symmetry in CAD Models,” ASME Design Automation Conference, Minneapolis, MN, Sept. 1994.
I recognized an application of this at home and thought the practical example might be useful. Please take a look at the picture, below. It is the cap of a child’s sippy-cup, which has two bosses inside it where a silicone rubber piece – which prevents leaks if the cup is tipped – is pressed into each of those two bosses (and is removed each time for washing). So here’s where partial symmetry comes in: the two female bosses on the cap’s inside are different diameters, which means the male parts of the silicone piece also need to be different diameters in order to fit snugly to prevent leaks. The picture shows my attempt to assemble them the wrong way – it doesn’t assemble. (Click on image for bigger picture.)
Each time I put this together I need to squint at the silicone piece to figure out which end is which; it takes time, and is getting on my nerves. The time to recognize which end is which is especially critical when my daughter is screaming “Juice! Daddy I want juice now!”
One thing I mention repeatedly when discussing DFA is that DFA is performance-ignorant. So I will grant you that there might be some functional reason why these two bosses might need to be different sizes. But it’s a mystery what that reason might be. I see no significant differences between the two sides of the insert, which has the stopping-leaks function.
During a design review, the question should have arisen: Why does this asymmetry exist? If there is no functional reason why it needs to be there – and, again, I see no reason why it does – the two bosses inside the cap and the silicone insert should have been redesigned so that the insert was functional in either orientation. However, if there truly was a need for the asymmetry, then the design should have been altered to make the asymmetry more distinctive.
Lesson One: When components are almost symmetrical, during design reviews ask whether the asymmetry exists for a functional reason. If there is a functional need for the asymmetry, design the parts to be less symmetrical so that the difference between orientations is clearer – even to the point of adding cosmetic-only features to help people differentiate which orientation is which. If there is no need for the asymmetry, then design the parts to be truly reversible.
A second lesson exists. This is not based on actual results, but merely my own observations and thoughts on symmetries and asymmetries. (If I ever did a PhD, I’d love to study this further.)
Lesson Two: If you need to put in an asymmetry, don’t just try to distort the shape or alter the size of something, as was done in this example of merely different diameters. Instead, add or subtract a feature so there is a feature-based difference between the orientations, not just a size-based difference. This, I believe – without empirical proof though it “makes sense” – will make that orientation requirement far more noticeable.
And a final observation about partial symmetries:
Lesson Three: If a component needs to be asymmetric, do your best to make sure that it cannot be assembled the wrong way. (I concede that was done here.)
Let me give an example of where this didn’t happen; this story is in my ASME paper, actually. A small felt wick was assembled into a high-speed bearing; its function was to wick oil from a reservoir to the bearing for lubrication. First, the wick was partially symmetric as it had a slight taper on one end, but the difference was small and complicated by the wick being a small part to begin with. Second, it could be installed in either orientation, and the people didn’t know. Needless to say, after a few months on the market, bearings began to fail left-right-and-center. Which leads to:
A corollary to Lesson Three – make sure your line operators know too. During my grad school research I came across an example of a yoke arm (link is an example only) whose mounting holes on either arm were partially symmetric. These two holes were just slightly different diameters to ensure proper orientation of the mating piece – this was done deliberately. However, since the operators didn’t know there was a difference, and there were no written work instructions detailing that there was a difference, the operators would attempt to assemble the unit and then find one hole was “too small”. They would then blame the machinists (doubtless with colorful language), and enlarge the smaller hole until things went together.
Lastly, I would be a poor criticizer if I didn’t attempt to show what I think might be a good redesign. So let me assume that there is, in fact, a functional reason why the two sides need to be different – please look at the other picture, of a CAD model I whipped up. (Click on image for bigger picture.)
- One side has small tabs, which should match with new notched cut-outs in the cap’s boss (following Lesson Two but also Lesson Three).
- The tabs add material, as do different sized fillets between the bosses and the center connecting piece; this is to attempt to balance material between the two sides so the gate can be in the center.
- The two bosses are different diameters, to ensure Lesson Three applies.
I had been downsized out of one full-time job in 2003, and it was before being hired full-time at another company – so I was doing contract work to “put beans on the table” and was working at Mack Molding as a project manager for a Lego toy truck. This was a fun project, and I’ll give credit to Mack Molding for letting me do the project plan the right way rather than throwing together a Gantt Chart a priori and then having that become holy writ. (In January when I mapped out all the tasks to create the project’s critical path I said we would be building the first production pieces the first week of July; right on schedule, the first week of July, the first production pieces rolled off the assembly line.)
Since this was a consumer product both safety and cost were critical concerns and I, as the project manager, had to choose the materials with that in mind… but child safety was paramount, not only for the materials but for the colors. So the very first criterion was to use FDA-approved resins and masterbatch colorants. Second was the resin cost, and we went with a polypropylene material for cost and because similar applications had also used this material – so there was some prior familiarity with it; as I recall it was a co-polymer polypropylene, for better impact resistance (see #6 at the link). Since Mack Molding had long-standing relationships with several resin suppliers, I leveraged their – and Lego’s – knowledge of prior resin and colorant use.
Plastics Lesson One: Leverage the experience of your suppliers to help you choose a suitable material. If you view them as a partner who has a stake in the success of your product, and make sure they know they have a stake in its success, they will be vested in making sure you get what you need to be successful.
Polypropylene is a polyolefin, and one of the characteristics of this kind of resin is that it has a wide range for shrinkage. Shrink is the term used for the fact that the part is designed to have specific dimensions; the mold then has to be cut to account for the fact that plastic – injected as a liquid, solidified, and cooled – shrinks some percentage. This shrinkage needs to be accounted for when cutting the steel for the mold. In the case of polypropylene, parts can shrink several percent (e.g., from 1%-3%) depending on any number of factors. So my first responsibility in dealing with the material selection, after selecting the material itself, was to choose the shrink rate to be used to scale the mold so that the parts come out right to the design dimensions.
So what did I choose? A low shrink rate at about 25% of the way from the low end of the window to the higher. Why? Because it’s easier to process a part to shrink more, rather than shrink less. For example, to have a part shrink more, you can reduce the force packing the plastic into the mold during injection, reducing the part density slightly – which increases shrink. One can also eject the part a little earlier from the mold, allowing it to cool in the air instead of in a hard fixture (the mold itself) which would hold it in place… in-air cooling allows the part to move more.
Conversely, to make the part shrink less, one can pack the part harder (increasing the material usage and cycle time) and hold the part in the mold longer (increasing cycle time). Both of these increase the part cost.
Plastics Lesson Two: When faced with a shrink rate window, choose towards the lower end rather than the higher; if the shrink rate needs to be massaged during trials to meet dimensional targets, it’s easier and cheaper to get the part to shrink more than to get it to shrink less. But don’t go hard-up against the low end of the range either as you want to leave a little wiggle room “just in case”.
But there was another layer to choosing the materials for the toy – the wheels needed to spin freely. And they needed to not only spin freely all the time, but potentially after sitting on a shelf a potentially-extended period. Also, the majority of drops would be onto the wheels, and while the main body components were made from a modified polypropylene, specifically for impact resistance, the wheels would undoubtedly bear the brunt of any impacts.
Something that people unfamiliar with plastics might not know is that similar-chemistry plastics sliding on each other have higher friction than two dissimilar plastics. Another datum for consideration is that like plastics can micro-weld to each other over time, though admittedly at room-temperature this is not likely. This immediately triggered my decision to make the wheels from a different resin type than the base product which, as stated above, was made from polypropylene.
I selected polyethylene as the material for the wheels (FDA approved, of course). First and foremost was the fact that it, too, is an inexpensive resin. It also has excellent impact resistance – better than polypropylene. And although still a polyolefin, its different chemistry would mean that the wheels would rotate freely and would not micro-weld to the body axles. The one wrinkle was that this would mildly complicate regrinding scrapped parts as the wheels would need to be segregated; however, since the colors would need to be segregated anyway, this was not an overly-burdensome hurdle.
Plastics Lesson Three: If products need to slide on each other reliably and with low friction, consider choosing dissimilar chemistries for the parts that need to move against another.
As I said early on, I did the project plan where we launched right on time. Part-and-parcel with that was the fact that in any project where parts have to be injection-molded, there really need to be at least three mold trials:
- Initial shots. Can plastic be put in, does the part come out close?
- Testing parts. Parts made for initial product testing, form-and-fit checks.
- Final verification before shipping. All changes made correctly?
Plastics Lesson Four: Budget time in the project plan for at least three mold trials. If there are critical aspects of form, fit, or function that are exacting in their requirements, you may need additional trials.
A project manager should not only be skilled in project management, but have a deep and instinctive understanding of the materials being used and the tasks required to bring the product to fruition. Just as a conductor is not just waving a wand to coordinate the orchestra, they must be a skilled musician in their own right to know the capabilities and limitations of the musicians, and instruments, under their direction.
© 2014, David Hunt, PE