For the TL;DR generation, this post describes using a 3D printed mold for making a complex fiberglass shape – in this case it’s an airbox for a Lycoming O-360 engine.

Fiberglass parts are often made using modes. For mass produced parts, molds are typically designed as ‘negatives’. The mold is hollow and the fiberglass is applied into it. When cured, the part is popped out and the mold is ready to be reused. For individual custom parts and for very complex shapes, the mold may be a ‘positive’ with the fiberglass applied to the outside. In this case, the mold material – typically sculpted foam – is sacrificed at the end when it is broken up and removed from the interior of the fiberglass part.

The majority of background I needed to know to accomplish this task came from two sources:

  • A valued contributor on the Vans Airforce forums – Dan Horton – wrote a series of posts detailing the use of foam molds for small fiberglass parts. I’ve saved his material as a PDF which is available here.
  • A builder of a Lancair 360 – Chris Zavatson – has a website dedicated to his aircraft and the work he did to improve it’s induction system. I’ve saved his material as a PDF which is available here.

Armed with Chris’s design ideas and Dan’s fabrication process, I embarked on creating an airbox which would provide enough air for the O-360 but fit in the available space of an RV-8 O-320 lower cowl. The critical dimensions were as follows:

  1. air filter size
  2. air inlet location
  3. carb heat mechanism
  4. distance between the O-360 carburetor and O-320 lower cowl
  5. size of the lower cowl ‘bump’ for the airbox

Of all of the dimensions, the most difficult was the fact the O-360 carburetor is nearly 2 inches taller and thus reduced the available distance for the airbox to only 2.5 inches thick below the carburetor.

I went through 5 iterations until I had the final usable airbox. I’m only going to describe the successful one.

The Mold


I had tried carving a foam block to make the mode. Ultimately, I could not maintain the very tight tolerances I needed to maximize the airbox while having it fit the available space. I chose to design the airbox in Autodesk Fusion 360 CAD software. Designing in CAD allowed me to create a shape which incorporated all of the critical dimensions.

The design required the ram air entering the forward inlet be slowed and expanded to pass through an air filter and then concentrated and accelerated only as much as needed to be channeled into the bottom of the carburetor. I chose to use the air filter recommended by Vans Aircraft for the IO-360 installation. My reasoning was simple – I wanted an air filter which would allow enough air for the O-360. Lacking any airflow engineering knowledge of my own, it was a safe bet their engineers had done the homework. This gave me three cross sectional data points – the cowl inlet size and location, the air filter size, and the carburetor inlet size and location.

The airbox design needed to maintain an airflow cross section greater than the inlet on the bottom of the carburetor. CAD allowed me to position the key cross sections and generate spline curves between them. The result was a compromise between optimal flow and available space. Working in software also allowed me to design the carb heat flap mechanism. In one position, the flap needed to completely close off the air inlet from the cowl, while in the other position, the flap needed to completely cover the hot air supply and maintain the interior shape of the airbox to minimize any disruption of the air flow. This was easy to accomplish and test in CAD.

I used Dan’s technique of creating ‘rails’ in the model. The air filter is trapped between these rails. The finished airbox has a seam between these rails which allows the airbox to the separated to perform cleaning or replacement of the filter. In the finished airbox – but not in the model – there is also a seam for the section under the carburetor. This allows access to the installation bolts which attach the airbox to the bottom of the carburetor.

I 3D printed the model and test fit it to the engine with the cowl. I made a few small adjustments to increase the clearance to the cowl to account for vibration and to allow for a rubber transition gasket between the cowl inlet the the airbox inlet.


I attempted to carve a foam block to match the 3D print but I was unable to maintain the necessary accuracy. I investigated the heat properties of fiberglass. My idea was to find a temperature at which a 3D printed plastic would be soft and flexible while the fiberglass object was still rigid.


PLA filament prints at 190C-215C. However, it becomes soft at a much lower temperature. It is easy to soften PLA using a heat gun. I heated a test print made from PLA filament to 150F and then 170F for one hour. The result was pliable while retaining its basic shape. It was also reasonable easy to handle for short periods of time.

I 3D printed the final mold. It did not fit on the print bed of my 3D printer so the model was split in CAD and the two parts were printed and then glued together. The mold was lightly sanded to remove most of the print lines and then coated with PVA mold release.

BTW: for those thinking I could have printed the model with PVA dissolvable filament, I tried that and it didn’t work. The PVA is too flexible and deformed during vacuum backing. HIPS could have been used for the model and then dissolved with d-limolene. However, I was not sure what effect the d-limolene would have on the fiberglass and the PLA heating method worked well.

Because of the model’s complexity, I chose to fiberglass it in two passes. The first pass started at the inlet and ended about 2/3rds up the face before the air filter. I created a filler of epoxy resin and flox and used in in the tight corners around the carb heat tube / flap area. While the filler was still wet, I then cut and applied three layers of fiberglass around the inlet, the flap transition section, the carb heat tube, and around all four sides of the transition to the air filter. I deliberately left an irregular finish in both position and number of layers because the second pass would overlay this first pass.


I applied peal-ply to the entire mold – even the bare section. I then wrapped the most complex areas in loosely wadded bleeder cloth before applying two full layers of bleeder cloth to the entire mold.

I had previously laid out my vacuum back. I deliberately made it longer than necessary. This allowed me to simply cut off the end after the first phase has cured and reuse the bag for the second phase.

For the second phase, I used Dan’s technique of filling the rails with a mix of epoxy resin and flox. While the filler was still wet, I pressed in a series of nutplates. (The final screws into the nutplates are shown in the finished photo at the start of this post.)

The second layup was simpler than the first, with larger areas and less complex transitions. I paid extra attention to the back top of the mode was critical and needed to be very flat and smooth because this would be bolted to the bottom of the carburetor. In the end – after cutting the hole in the airbox where it meets the carburetor inlet – a layer of carbon fiber was added to increase rigidity.

Once the entire mold was fiberglassed, the location for the seams were drawn and cut approximately 1 inch deep using a thin cutt-off wheel. Dan describes the technique for making the seams. Packing tape is applied to one side of each seam and then 2 layers of fiberglass strips are applied with half the width on the exposed fiberglass and half the width over the packing tape. (If you haven’t figured it out by now, you really should read Dan’s PDF.)


I placed the airbox in the oven set to 170F and used a thermometer to verify it did not over heat. I left it for one hour. At the end of the hour, I was able to easily remove the rear bottom fiberglass part and pull out the mold which separated at the glue joint (from the original two 3D printed pieces). That left a large opening in the back and a combination of pushing the pliable mold  down from the hot air tube and pushing it in from the inlet, the remainder of the mold was pulled out the back.


The finishing tasks included screw holes, adding the nutplates for the rear seam, fabricating the carb heat flap, and attaching the rubber gasket to the inlet.

The translucent fiberglass made it easy to locate and drill the holes to match up to the nutplates. A holes for the rear seam were drilled and then screws were temporarily inserted into the upper second of the airbox and nutplates attached. Then a mix of epoxy and flox was created and a fillet formed to contain the nutplates and act as a bit of transition to help direct the airflow up to the bottom of the carburetor.

The hole was cut to match the inlet shape of the carburetor. The airbox felt it should be more rigid and a layer of carbon fiber was added and then trimmed.


The surface of the fiberglass was still quite rough and uneven. This was mostly from poor technique and insufficient attention when applying the peel-ply and bleeder cloth. The surface was lightly sanded, avoiding any damage to the fiberglass cloth. The majority of smoothing was accomplished by creating a mix of epoxy and micro-balloons – the consistency of cake frosting. This was applied with a flexible plastic squeegee. The hardened filler was sanded and then covered with two coats of filler primer. The primer was sanded and painted with two coats of finish paint.

The airbox was again test fit to the carburetor and the cowl. The screw holes are easily visible along the seams. The seam for the air filter has a total of 8 screws and nutplates. The seam for access to the installation bolts has 14 nutplates.


Once the airbox was finished and the carb heat flap installed, the finished assembly was attached to the carburetor. the carb heat control cable was attached and the lower cowl was fitted … well, attempted.

It turns out the design insured the airbox fit but the control cable for the carb heat didn’t have sufficient clearance in the troth of the lower cowl. The cable was relocated further up and a bell crank was designed to redirect the control force. (The photo at the beginning of this post shows the working bell crank.)

Throughout the design process, the solution was evaluated for failure modes. The carb heat mechanism was given special scrutiny. In the event that any potion of the mechanism should jam or break, the engine will continue to get air from one or both sources. Additionally, the air filter would prevent any FOD from reaching the carburetor.


The airbox has 5 hours of operation and thus far is performing well. Static RPM is as predicted by the propeller manufacturer and a ground test with and without the air filter installed demonstrates there is sufficient air is reaching the carburetor inlet through the air filter. Once engine break-in has been completed, a long endurance flight will be conducted to validate the airbox under heat-soaked conditions.

Ultimately, I’d like to duplicate this airbox using all carbon fiber. The only challenge will be locating the embedded nutplates to drill screw holes. With the fiberglass, the nutplates are visible. The likely solution will be to install ‘indicators’ so the nutplate locates are obvious under the completely opaque carbon fiber. I’d also like to make another airbox to hone my skills and prove I can do a better job with a better finish.