Final Project: The Pokeball chess-Allen, Joe, Louis & Will | ENGI 210: Prototyping and Fabrication

PART I: Introduction

This is our final project. Honestly, I felt overwhelmed when I received the instructions at the beginning. Joe and I had just finished the midterm project and expected to have more time to relax. However, the good news was that we had two additional partners, Louis and Will, on our team, which meant we had more hands and minds to collaborate.

This final project integrates 3D printing, CNC machining, molding, and casting to create four Pokeball chess pieces. We also became familiar with operating machines using the following software: Meshmixer, Carbide Create, SOLIDWORKS, and VCarve. It has been an impressive learning experience.

PART II: Workflow

This project was divided into four parts: Gate 1, Gate 2, Gate 3, and Gate 4. Joe, Louis, Will, and I attempted to complete this project step by step.

Gate 1: Design choosing
We chose Pokeball chess from the Thingiverse website for our final project. This design was created by Roshandp1 on April 11, 2016. After acquiring the file, we scaled its dimensions to 38.397 x 60 x 38.409 mm.

↑↑↑ The object made by 3D printing
.
※ [Issue Gate 1-1] ※
As part of the final project requirements, we need to utilize CNC machining and 3D printing to create the positive mold. To accommodate the limitations of the 3-axis CNC machine, we must avoid any designs that include overhangs. Initially, we chose a different design; however, we noticed that the top portion of that object featured an overhang. As a result, we decided to switch to the design of a Pokeball instead.

↑↑↑ The overhangs part on the left ↑↑↑ The overhangs part on the top
.
Gate 2: 3D print positive mold
We used SOLIDWORKS and Meshmixer to split the object in half and added a plate structure to the bottom. Since one side of the Pokéball has a more complex design than the other, we decided to use 3D printing to create a positive mold of the intricate part, as shown in the picture below. Next, we applied cardboard along the entire 3D-printed positive mold. Finally, we used silicone to create the negative mold and de-mold overnight.

↑↑↑ 3D-printed positive mold
.

↑↑↑ Silicone negative mold
.

↑↑↑ Silicone negative mold
.
※ [Issue Gate 2-1] ※
When we used cardboard to create a box around the 3D-printed positive mold, we applied hot glue to secure it. To prevent any leaks while using silicone to make a negative mold, we sealed both the outside and inside of the cardboard with hot glue. However, after the teaching assistant explained the process, we realized that we should not have added hot glue inside the small cardboard box, as it could damage the details of the object. Consequently, when we worked on the cardboard for the CNC part, we corrected this mistake and achieved a better negative mold.
.
Gate 3: CNC print positive mold
In this step, we aimed to utilize CNC technology to create the second half of the positive molds. We started by using SOLIDWORKS to add a plate structure to the bottom of the object. Next, we designed the CNC toolpaths using VCarve. The CNC machining process consisted of two steps. First, we used a 0.25-inch end mill to create the rough machining toolpath, which took approximately 19 minutes. In the second step, we designed the finish machining toolpath using a 0.125-inch ball nose, which took about 40 minutes. After completing the CNC processing, we coated two layers of lubricant on the surface of the wood. Later, we applied cardboard along the CNC positive mold and used silicone to create the negative mold.

↑↑↑ Toolpaths design
.

↑↑↑ 0.25-inch end mill ↑↑↑ 0.125-inch ball nose
.

↑↑↑ The small box was made of cardboard, and hot glue was used to attach it to the wood.
.

.
※ [Issue Gate 3-1] ※
The piece of wood we received was larger than the last one, which made it difficult to secure it on the bed of the CNC machine. After Karl suggested an adjustment, we drilled four holes into the sides of the wood so that the clamps could fit properly. Eventually, we were able to secure the wood firmly on the bed.

.
※ [Issue Gate 3-2] ※
During our initial experiences with VCarve, we spent considerable time designing toolpaths. We understood that toolpath design consists of two main components: rough machining toolpaths and finish machining toolpaths. However, we initially only used rough machining mode to design the toolpaths. Our original design indicated that the process would take 88 hours or 612 hours for rough machining. After discussing this with Luke, I recognized and corrected the error in my design. Ultimately, we completed the entire CNC process under 1 hour.

↑↑↑ Original ↑↑↑ After correction
.
※ [Issue Gate 3-3] ※
During the lubricant coating process, we applied too much lubricant on the first layer to ensure good coverage on the wood. Typically, the lubricant should dry at room temperature in 3 to 5 minutes. However, we waited for over 10 minutes before proceeding. We then used KIMWIPES to remove the excess lubricant and applied a second, thinner layer. After 5 minutes, this layer dried completely at room temperature. We will remember the spray painting skills!!!
.
Gate 4: Polyurethane casting and post-processing
We combined the two molds and used rubber to fix the two parts together. Next, we prepared polyurethane by mixing solutions A and B according to the object’s size. Once solution A was mixed with solution B, we poured the mixture into the mold within 30 seconds and allowed it to polymerize at room temperature for 10 minutes. Afterward, we removed the Pokeball chess piece from the mold and repeated the casting three more times, as we needed to create four identical pieces for the final project. Finally, we used a graver to remove excess polyurethane and sandpaper for post-processing.

↑↑↑ The process of casting by using polyurethane
.

↑↑↑
The three pieces on the left were cast using polyurethane, while the one on the right is our reference made by 3D printing.
.
※ [Issue Gate 4-1] ※
Theoretically, we used the same design to create the 3D printing molds and the CNC positive molds, which means that the size and shape of the two negative molds should be identical. However, when we combined the two parts, we noticed that the middle areas were not correctly aligned. To resolve this issue, we dedicated additional time to post-processing to ensure the quality of each piece.
.
※ [Issue Gate 4-2] ※
We initially estimated the volume needed for our polyurethane using a 3D printing reference. However, when we compared our volume with another group’s, we realized we had overestimated it. To address this issue and avoid wasting polyurethane, we reduced the volume by half based on the size of the other group’s pieces. We continued to adjust the volume as we analyzed the remaining material, gradually reducing the amount of polyurethane until we determined the optimal volume needed for casting our pieces, which was 15 ml for either solution A or B.
.

PART III: The WorkSpace

The space was cleaned after used.

PART IV: Final Piece Overview

While we know that we were only allowed to cast 4 times, we really would have liked to refine our production process and create more pieces. Yes, we have quality pieces. Yes, we are proud of them. However, to improve and make future strides in molding, it is important to reflect on the shortcomings of our process. So, to dive deep, we had 4 iterations of the molding. We used many thin rubber bands in our first attempt at pouring the design. We likely used 10 thin (1 cm wide) rubber bands. When we took out the design from this setup, we noticed that our Pokeball came out a little warped, looking extra thin. This first setup clearly applied too much tension to the areas where the rubber bands were. To change this, we used one big rubber band (for the rest of the casting). This worked! Our pieces came out much more normal to their expected proportions. This said, we found that we had some more issues to address. To be honest, our mold didn’t line up too well and we messed up our lined up markers in the design, such that both male ends of the lined up markers touched, and both female ends touched. This made our molding process a little challenging, but we found that we could still use the correctly placed markers to line up our piece, though they had little holding power past that point. In turn, our secondary chess piece came out a little misaligned. With the help of an exacto-knife and sandpaper, we fixed this. In our third and fourth chess pieces, we think we really improved. We used the previously found data to make small corrections and get better-aligned pieces, squeezing with the rubber band and looking carefully inside the mold to check for alignment. Overall, we are very proud of our work, but we definitely would change a few things about our process. Mistakes will be made, and that is okay. However, we will use more even pressure in future silicone mold processes and also pay closer attention to alignment.

PART V: Cost Estimation

Cost Type	Cost	Price	Source	Quantity	Total
Materials	Unfinished Natural Poplar Wood Board (1.5″ x 30″ x 48″)	$75.98	lowes.com	1.7% board	$1.29
	PLA Basic Filament	$22.99	bambulab.com	6.6% pack	$1.52
	Silicone Mold	$9.99	amazon.com	104% bottle	$10.4
	Polyurethane	$40.99	amazon.com	4.2% tank	$1.73
	Fine 180-Grit Sheet Sandpaper	$6.68	homedepot.com	5.5% pack	$0.37
	Mold Release Spray	$16.99	homedepot.com	0.69% bottle	$0.12
	Popsicle Craft Sticks	$4.99	amazon.com	7.5% pack	$0.37
	Hot Glue Sticks	$8.99	amazon.com	4% pack	$0.36
	Disposable Plastic Measuring Cup	$14.99	amazon.com	10% pack	$1.49
	Total Material Cost: $17.65
Labor	Molding Operator	$16/hr	ZipRecruiter.com	2 hr	$32
Labor	Prototyping Engineer (You!)	$25/hr	Indeed.com (Engineering Intern)	1 hr	$ 25
	Total Labor Cost: $57
Overhead	Facility Cost	$36/hr	plasticstoday.com	2 hr	$ 72
	Quality Control	$22.5/hr	Glassdoor.com (Quality Assurance Inspector)	0.5 hr	$11.25
	Total Overhead Cost: $83.25
Design	Engineering and Development	$25/hr	Indeed.com (Engineering Intern)	5 hr	$125
Total					$282.9

Total: 282.9

The total cost of manufacturing the four Pokeball chess pieces using molding techniques is $282.9, meaning one piece was about $71. That is quite the price for a chess set (not unheard of) but a lot for an individual piece. Notably, the majority of the expenses- nearly 93% of the total cost- are attributed to labor, overhead, and design. Among these, the five-hour design process for engineering and development alone represents 44% of the total manufacturing cost. Regarding material costs, the silicone mold constitutes almost 60% of the total material expenditure. Clearly, there are parts of our process that bias the final cost of our piece. How do we change this? Well, we can definitely get more advanced in the process. With additional skill or automation, we could reduce the cost we spend on manufacturing. Likewise, we could reduce the cost of our part by buying material in bulk and minimizing the number of mistakes (waste) in our materials. Reasonably speaking, one of our pieces was warped and would be flagged for quality control. Thus, our pieces are even more expensive (~$94).

Let’s talk business models and scaling our production for profit. We will settle with a mass manufacturing model. Within this model, we will establish partnerships with large corporations like Pokemon. We believe that this established partnership (or partnerships similar) will allow for years of stable production and for our sets to begin to generate their own brand by profiting from our partners consumers. Making plastic chess pieces in iterations will become more expensive if we make the whole chess set–there 6 pieces in a set (knight, king, queen, bishop, rook, and pawn)–so we want to make one set of molds. This is a greater reason to stick with one key, strong partnership, like Pokemon, for our first set of pieces.

Looking ahead to scaling up the production, we would have high initial costs for labor for overhead, design, and mold materials (wood and silicon). This said, the potential for reuse of the silicon mold could significantly lower the cost per piece in future manufacturing runs. We estimate that we could functionally make one manufacturing batch of 100 based on a single silicon mold–we saw minimal wear and tear in our first batch of pieces. With this model, we roughly mitigate the biased costs of our existing chess piece. While it is hard to estimate the exact price due to changes in both the overall product and model, we are confident in our thought process behind this model.