Tag Archives: 3d printing

3D Printing and STEM Education

As the Maker Movement spreads into the halls of nearly every school across the country, and with it the technologies that tend to be synonymous with that movement, I thought it might be useful for me to write a reflection about how we have used 3D printing in our program and some of the things that others might want to consider when thinking about investing in 3D printing for their school.

What is 3D Printing?

A 3D printer is any device that uses additive fabrication to essentially create some three dimensional object by building that object layer by layer. Currently the most popular way to do this in the educational sphere (because it is the least expensive) is to build the models by extruding melted plastic – similar to having a very precise hot glue gun. This has been the technology that we have used for the past five or six years in the Academy. Recently there has been an explosion in new inexpensive models coming to market that use light to harden photosensitive liquid polymers. These technologies (known as stereolithography  and digital light processing – DLP) use a focused light source or laser to harden the liquid layer by layer. The finished model in most cases is made of some kind of plastic – ABS, PLA, etc. Although the technology is moving forward with other materials, any kind of printer that would be used in a classroom environment is going to make plastic models.

Whatever the process of the 3D printer, these technologies are different from subtractive manufacturing which starts with a “chunk of stuff” and then carves the material away, leaving the 3D model. These always require some kind of cutting instrument like a hardened metal drill bit, or even possibly a laser or high pressure water jet. These methods are still preferred for the actual manufacturing of things like airplane parts, high precision medical instruments and all sorts of other machinery because these methods are highly precise and can be used to create parts from almost any material – metal, stone, plastic, etc. There are two issues with subtractive manufacturing though. The machinery is generally very expensive and learning how to use it properly is quite challenging.

If you would like to learn more about 3D printing in general, I suggest this great website:

http://3dprintingindustry.com/3d-printing-basics-free-beginners-guide/processes/

Our Printers

We currently have three operational 3D printers and one new DLP printer that we are currently assembling. The first 3D printer that we purchased was a Stratasys uPrint:

This printer has served us very well. It builds very precise models using really solid ABS plastic. Its precision comes at a cost though – it is quite slow. OK, its really slow! A nose cone for a rocket can take upwards of six hours to build! The other drawback of this printer is the cost and availability of build/support material. We just bought a complete restock of material and it cost us nearly $1700! Keep in mind that this should last us about six years to eight years.

The other two models that we have are the ubiquitous MakerBot Replicator 2’s. These are much simpler to operate (when they aren’t clogged) and they are much faster. They are also much less expensive. The uPrint cost us about $20,000 dollars including the rinse tank, while each MakerBot Replicator 2 cost us about $2200. Actually one of the MakerBots was part of a DonorsChoose/Autodesk program that cost us nothing (thank you donors and Autodesk!). The material for these machines is much cheaper – about $90 to $50 dollars per spool, as opposed to about $200+ per spool for the uPrint. The drawbacks of these machines is that they need constant maintenance, manual calibration, and the models that they build are not as accurate nor as precise.

We recently were very honored to be the recipients of a new 3D printer, donated by our high school’s parent organization – WeAreSR. Although we haven’t yet been able to use our new 3D printer from Kudo3D, we are excited by its potential. This DLP printer is said to have a much higher resolution, a much faster build speed, and a very large build volume. We will be posting an update once we get it running – which should be soon!

Rapid Prototyping = Rapid Learning

The really rewarding educational aspect of 3D printing, from a teacher’s perspective, is the acceleration of the learning cycle. Students can quickly identify weaknesses in their designs because they can have a part in their hands in literally hours, then make adjustments and have a new version fabricated, sometimes in a single class period. This would be nearly impossible ten years ago.

Now some might argue that it relieves students from the importance of having to think more carefully about their work, but I think this is outweighed by the advantage of allowing students to more quickly assess their spatial reasoning, and as long as we teachers force the students to also reflect on why their design failed or needed revision, then I think ultimately the students will learn more quickly.

This does not mean that we always allow the students to print whatever they want. We do act as “gatekeepers” to the printers so that we aren’t wasting student time, our time and resources. The models must pass a few minimal requirements, such as double checking dimensions, seeing if the model could be made more efficiently as multiple parts, etc.

3D Printing = 3D Spacial Problem Solving

The 3D printers have acted as a great arena for students to learn and develop their 3D spatial problem solving skills. To be clear, its actually the combination of 3D CAD software used to design the models and 3D printing to create the actualized models that helps students visualize, navigate and anticipate interesting three dimensional problems. Generally, the printers are used to create parts that are then used in more complex assemblies. The interface of these parts is where we see students encountering and having to solve complex spatial puzzles.

One of the things that I have witnessed is the advancing complexities of student designs as they become more familiar with the software and also develop their ability to mentally construct the spatial relationships between assembly components. At some point, I’d like to document this process and perhaps develop an assessment tool for measuring the development of these cognitive skills.

The Limitations (Not Star Trek Yet…)

The really amazing aspect of 3D printing is the ability to create real objects from imaginary ones with an almost perfect translation. I do think that it is important to realize that there are some limitations and also some things to consider before you run out and buy one of these things. Here are few things to consider:

All Those Plastic Things

One of my biggest complaints to the 3D printing industry is the lack of any clear and clean way to take 3D printed models that were unsuccessful and break them down back into raw materials for use in the printer. At the end of the school year, we have a fairly large bin of unwanted models that we collect for recycling. Some of the models are indeed recyclable while others are not depending on the material used. I think the manufacturers need to come up with a clear “cradle to cradle” solution for their printers that allow users to throw their models back into the machine to be re-extruded. It is theoretically possible and at least one company is offering a product called the Fillabot for addressing this issue.

Its Not That “Rapid”

Now, in relative terms when compared to milling, 3D printing is pretty fast, but it is actually slow in the context of the classroom. Even though its called rapid prototyping, it can seem really slow for some folks who are new to the world of manufacturing prototypes. You see, in the past, modelers would make a prototype out of clay, create a mold, cast the mold, make refinements, etc. Or one would calibrate and setup the CNC mill, have to change out bits, run test cuts, etc. In this context, 3D printing seems rapid. But its still not Star Trek.

The time can vary significantly based on the type of the printer, the complexity of the design, and the size of the model. This can be really frustrating for some teachers who want to be able to print an entire class’ models and have them ready for the next class period – that won’t happen. It can take hours to print just one model. You have to design your course in a way that allows the students to work on parallel tasks and then you need to have some way of keeping track of the printing queue.

It Won’t Make Everything

These machines are amazing, and we really love our array of 3D Printers, but experience has taught us that there are limitations to what they can make. Because all of these printers essentially work with liquified plastic, there are limits to the geometry of what you can build. As the models are built, they can “sag” or deform under their own weight. This can lead to small deformities, or catastrophic failures. Calibration can also be an issue with some of the less expensive or older printers. If the build plate is not properly leveled and calibrated, the entire build process can fail. Models with significant “overhang” can collapse, ruining the model.

Different printers deal with this slightly differently. Our MakerBots, for example, add “supports” to the model. These are little posts that act to hold up arching forms. The problem with this is that these posts then need to be removed from the model, and we have found this to be less than ideal. It adds extra time to the process because you have to do some post finishing work which can include filing and some sanding. Our uPrint actually adds a soluble support to the model that can be removed using a mild (but still toxic) solution. Again, this post processing adds a significant amount of time to the entire fabrication process.

Size is also a limitation. Don’t think for a minute that just because the build plate is 8 inches by 6 inches, that you can build a model with that footprint. You can’t. Once again, because you are dealing with liquified plastic that cools, it also shrinks. The larger the volume of the model, the greater the chance that the model will curl, buckle, and deform. Read the fine print from the manufacturer to get the real build size limit.

Easier, But Not Easy

The last point we want to make is that working with a 3D printing is certainly easier than running a 5 axis CNC mill, but they are not as easy to work with as an actual 2D printer. Adding that extra dimension has its challenges. Plan on spending quite a bit of time learning how to maintain your printer. Just like 2D printers, 3D printers “jam” all the time. The extruded plastic can get stuck in the nozzle and you can come back after several hours of printing and find that the very last cm of the model never printed because the nozzle is completely gummed up! Be aware that these can be infuriating moments that take significant amounts of time to fix. I have spoken to some teachers that got so frustrated that their printers ended up just sitting in a corner of the classroom, tragically unused.

Our Recommendations

Our recommendations are simple. Before you go out and buy one of these things, you have to be willing to put in quite a bit of time to maintain it and learn how to optimize your printer’s performance. There are tricks to optimizing each printer out there, and it will require that you watch some YouTube videos, dig through online support forums and be patient.

There are clear and obvious reasons to get one of these if you are running a STEM program, especially one focused on engineering or design. What might not be obvious is that these machines can also be incorporated into mathematics education, and definitely into a 3D art course or sculpture course.

There are so many models out there now, and they all claim to be the very best value. Each will obviously have advantages and disadvantages. Ease of use and less expensive generally means that your models will not be as precise or accurate. Inexpensive models can also be difficult to maintain. DLP printers are looking promising. They are coming down in price, they are faster and they are very precise. They can print models using different materials (like castable resin, or flexible resin). On the other hand, keep in mind that they are still more expensive, and they use a somewhat toxic resin that can be a non starter for some teachers.

We Have Lift Off!

15414703973_b0258a08ca_h  15846865828_b08d8515cd_h 16033503742_be978fd9ca_h

This is a follow up post to Modeling A Rocket’s Journey – A Synthesis where I described how the students in the first year program were engaged in creating a predictive report for their model rockets. I want to emphasize that these model rockets were not kits. Each rocket was designed using 3D CAD software, and each component was either fabricated from raw material, or was created from material that was not intended for use in model rocketry. The only exception to this is the actual rocket motor.

The next step was to launch the rockets and have the altimeter payload collect altitude data.

Launch Conditions – A Bit Soggy

Unfortunately the week of our scheduled launch happened to be a week of some pretty hefty rains. We rescheduled the launch twice before finally accepting the soggy launch conditions. With umbrellas and rain jackets, we trudged out to the baseball diamond and got to work setting up for the launch. We had some minor difficulties in the wet weather, but eventually had a very successful launch day.

Most of the rockets were able to launch and deploy their valuable payload – the Pnut Altimeter.

15414689283_8197c70584_h

The students seemed very excited to finally see the rockets launch, and to see the successful deployment of the parachutes. Although we all got a little wet and muddy, we had a great time!

The Altimeter Data

The altimeters use a small barometric pressure sensor to collect altitude data (the altimeters also contain a small temperature sensor and voltage sensor). The altitude is recorded in feet every .05 seconds. Here is an example of one rocket’s recorded flight data:

altitude_vs_time

https://plot.ly/~stemples/9

The students were then asked to use the data to create a comparative analysis report. I will detail how the assignment was set up and also discuss how the students performed on this assignment. That will be for another post.

I want to also thank Mr. Kainz for his amazing photos that are displayed here.

Building The Net Torque Model – Part 3

Appending The Conservative Models

After investigating the causal relationship between torque and angular acceleration, I introduced the possibility to the class that perhaps we also needed to revisit the Energy Transfer Model and the Momentum Transfer Model. The students agreed that an object that is rotating must have energy. This was pretty easy to demonstrate.

I set up a situation in the class where two of the variable inertia disks that we created on the 3D printers were placed at the top of an inclined ramp. The internal marbles were placed at two different configurations inside the disks and then the students predicted which disk would reach the end of the ramp first. I was pleased to find out that the class appeared to agree that the disk with the marbles located closer to the radius would be the winner. I really think that our investigation with the variable inertia disks solidified the students’ conceptual understanding of rotational inertia and the importance of mass distribution.

I have not yet found a good experiment where students could discover the rotational kinetic energy relationship, so I decided to take them through a derivation based on linear kinetic energy. I then asked the students to do some whiteboard work. I asked them to demonstrate that the disks would indeed reach the end of the ramp at different times. Although this wasn’t strictly a constructivist approach, it was good practice in doing some fairly difficult algebra without numerical values – something the students traditionally are not very good at.

We then moved onto momentum. Again, I started by reviewing the Momentum Transfer Model for a particle. At this point the pattern had been fairly well established. The relationship between angular and linear quantities seemed to have taken hold because the students were quick to propose a mathematical definition for angular momentum. Our next goal was to figure out whether this was a conserved quantity.

Mr Holt and I had created a set of metal disks that could be attached to the rotary sensors. I decided to create our own, rather than (sorry Vernier) buy them as I thought that the commercial kit was over priced. It wasn’t too hard to create the disks, especially when you have access to a CNC plasma cutter!

image1

The students attached one disk to the rotary motion sensor and then got that disk spinning. They then took a second disk that had small magnets attached to it, and dropped this disk onto the spinning disk. The students compared the angular velocity before and after the disks were combined and then calculated the angular momentum of the system before and after. The data we got was quite good with the class getting in the range of only about a 5% to 6% difference.

Wrapping it Up (or Un-Rolling It Down)

As a final deployment, I decided to try the deployment activity that Frank Nochese did with his students. It seemed like a good (and fun) way to wrap up our model (or as I have already argued – models).

Before doing the deployment activity, I reviewed all the model specifics with the students. My point here was to impress on them that what we had not really built a new model, but rather had extended many of the prior particle models to include rigid extended bodies. This generally only required that we consider the moment arm in all the particle models. I think this really helped a number of students see the connection between models that they felt they understood and all this rotational stuff that seemed a bit confusing.

I then set them up with the deployment activity, but I asked them to specifically solve the problem using both energy and net torque. There was some success, but I realized that the task was a bit much for the class. Once again, it is clear that I need to give them more practice with these problems that require multiple steps and that involve algebraic manipulation of symbols without numbers. Plenty of time to practice that!

Building The Net Torque Model – Part 2

Investigating An Unbalanced Net Torque

We started by looking at the fact that a disk experiencing a net unbalanced torque also experienced a change in rotational or angular velocity. The students used a rotary motion sensor to measure the angular position and the angular velocity of a disk experiencing a constant torque and the students immediately recognized the similarity between a particle experiencing constant linear acceleration and a rigid body experiencing constant angular acceleration.

IMG_0704

So, the natural next step in our investigation was to determine the causal relationship between a net unbalanced torque and the angular motion state of a rigid body. The students discussed how we might set up an investigation that would help us understand this relationship, and I helped guide them towards a final investigation design where we used a rotary motion sensor attached to the wooden disk we had used in a previous investigation.

The rotary sensors from Vernier are rather expensive, but also quite nice. They come with a plastic spindle with three different pulley radii. The only issue that I have with these is that they do not include a useful screw for attaching objects to the pulley – they expect you to purchase their completely over priced accessory kit. I don’t suggest this. Instead, you can create your own and then use a 6/32 screw to attach them to the pulley – just be careful not strip the threads!

IMG_0703

Working off a similar investigation when studying the causal relationship between a net unbalanced force and linear acceleration, the students recognized that we needed to include a force sensor for measuring the applied force. The students then had LoggerPro plot the points for the calculated torque (using a calculated column) and the measured angular acceleration.  The data isn’t super clean, but its good enough for the students to conclude that the relationship is most likely linear.

Dimensional Analysis – Inferring Rotational Inertia

I was amazed to discover that some students in the class set off to understand what the units of the slope could be reduced to. They immediately saw that the proportional constant (slope) included kg (mass) but that wasn’t all. After some work, the class had determined that the units for the slope were kg * m^2.

This led to a qualitative discussion about the inertia of rotating objects. We discussed hoops and disks primarily, and the class seemed to agree that the units made sense. Although I didn’t have any hoop-disk sets like the ones you can buy from various vendors, we did perform some thought experiments around mass distribution and rotation, but we needed to be sure that we were on the right track.

A Better Variable Inertia Disk

So last year I was looking for an investigation that would really help the students discover the importance of mass distribution for a rotating object. Reading the material on the AMTA website regarding the unit on rotational motion, the researches stressed the importance of connecting mass distribution to the rotational inertia. I found this variable inertia disk from Fischer Scientific and decided to purchase it. I commend these guys on making this, but frankly I decided that I could make a better one.

My colleague and I set about redesigning these disks. The improvements we made included a) better compartments for the metal marbles so that they didn’t move around, b) an index and lip so that the two sides fit more securely together, and c) more compartments so that we can test more mass configurations. After we made our designs, we used our Makerbot 3D printer to print out ten copies of the disks.

IMG_0707 IMG_0708

With these disks attached to the rotary motion sensors, the students were able to confirm that although the total mass of the disk did not change, the angular acceleration declined as the mass was moved outward from the radius of rotation. This was more of a qualitative investigation, but I think the students were able to clearly see how moving the mass farther from the axis affected the rotational inertia of the disk.

Deploying The Model (So far)

Now that the students had built a predictive model that described the quantitative relationship between rotational inertia, angular acceleration and net torque, they were ready to test it.

We returned to the investigation setup with the disk, but this time attached a hanging mass to the string and they attempted to predict the angular acceleration. The students immediately made the mistake (as I thought they might) in considering only the disk in the system. This gave me a chance to review with them the following process:

  1. Identify the system schema (now it includes things that can rotate!)
  2. Draw your force vector (free body/particle) diagrams and now also your force-moment arm (rigid body) diagrams.
  3. Write the summation of forces AND the summation of torques for the system.
  4. Do some algebra.

The trickiest part to figuring this out is getting the signs right due to interaction pairs and making sure that they agree. I ask the students to start with a diagram and then go through and label each force + or – by picking a force and then finding its partner and then making sure that the rest of the force directions agree.

Once they were able to get a prediction that we all agreed seemed correct, they ran the experiment. The class was able to predict the angular acceleration within about 10% error. We discussed the possible reason for discrepancy which led to some interesting discussions around modeling the rotational inertia of the attachment screw and washer, and also the frictional forces at work.

The students were pretty convinced that the model worked, and so next stop – energy and momentum in systems with rotational motion.