Thermal analysis of a 3D printer Hot-end
3D Printing is a small market which tends to grow each year as enthusiast are becoming more en more informed and skilled. Companies and Start-ups continuously improve both hardware and software parts of the 3D printers and their accessories.
As I watch this technological bubble getting better and more recognized by the neophytes, I am always thinking of what could be the next step, what could be the next hardware improvement or the next industrial application. Understanding little by little how 3D printers were working (especially FDM printers), I was trying to find the weaknesses of the different subsystems.
Working on numerical thermal analysis (CFD) a few months back, I thought about the hot-end and its thermal behavior. The hot-end, is the part that melts the polymer filament and lays it on the printing bed.
It has to be very hot (220°C-280°C) but should not melt the filament too far up in order to allow a good flow of material coming from the step motor. The heat break stops the majority of the conduction heat flow from the heat block while the heat sink takes the last calories off the tube guiding the filament. This way the filament is only melted in the heat break, the heat block and the nozzle allowing a good filament providing and printing.
I wanted to investigate how efficient the cooling was and how it could be improved. Is a fan mandatory ? Is the duct really efficient ? What technology could we use instead ?
So I decided to run a CFD analysis on a E3D hot end type. This kind of hot end is one of the most used in the FDM world, on Prusa models for example.
The following design and simulation are made with SolidWorks Flow Simulation 2016.
This simulation features 3 different configurations :
- A : hot end without fan (not designed for use without a fan)
- B : hot end with fan
- C : hot end with fan and duct
The design comes from GrabCad.com as I did not have an E3D hot end to take measurements on. The design came in a .part document and I was able to modify some dimensions like the Heat Break wall thickness or the heat block height. I was also able to add a fan that is known by Solidworks Flow Simulation, hence facilitating data input.
Data and boundary Conditions :
- Static simulation (non time-dependent)
- External analysis
- Temperature of the air : 25°C
- Temperature of heating component : 220°C
- Atmospheric pressure : 101325 Pa
- Objectives : Max Temperatures / Max Speed
- Gravity acceleration : -9.81 m/s² (Z axis)
- Fan specs (when used) :
Model : 109P0412H902 (40x40x10)
Speed of fan : 649 rad/s
Air flow function of static pressure
The meshing of the hot end was quite straight forward, focusing on the fins of the heat sink, however, I was not able to reach the level of quality I wanted due to the poor performance of the computer I was using. For this application and the low speeds reached during the simulation, this level of meshing is reliable. I used a local mesh to increase the precision around the hot end and especially the fins of the heat sink. Three to five cells between small gaps is usually enough especially at these low speeds. The mesh is the same for the three simulations and provides accurate results.
1. Hot End without fan
Running the simulation on the entire design took a relatively short time (regarding computer capacities). I could have split the model in half but as it is not entirely symmetrical I preferred to keep it this way and have the full information.
This type of hot end is not supposed to be heated without a fan, but it seemed interesting to see how the hot end and the PLA would behave under such circumstances. It could also help us understand how we could improve or create a cooling system.
We can observe on the picture on the left that the temperature rises up to 120-130°C on the skin of the heat sink. PLA starts to melt at around 140-150°C and is already very flexible when higher than 100°C, thus, as the filament is pushed through the hot end, it could result in a PLA build up transmitting heat by conduction up in the heat sink and potentially blocking the heat sink when the hot end cools down.
The two pictures show that hot air is going up through the heat sink and thus warming it by convection. The second picture tells us that the hot air is creating a barrier preventing the good circulation of fluid between the fins, indeed the the movement of the air associated to the vertical geometry of the hot end makes the velocity of the air between fins close to zero preventing a good cooling.
We can see that the heat break works pretty well as the temperature drop between the closest surfaces of the heat block and the heat sink (around 6mm distant) is almost 110°C. As the heat break forms a thin bridge between the heat block and the heat sink and stainless steel has a relatively low thermal conductivity, heat is mostly transferred through air natural convection.
Regarding PLA, which is really the aim of the simulation, we see that it mostly follows the temperature pattern of the hot end. The PLA filament reaches 140°C at the bottom of the heat sink, theoretically below the melting the temperature, and so allows to behave properly as it only melts in the expected area.
However, Solidworks Flow Simulation does not support melting materials, meaning that is does not predict how the material is going to melt and how it is going to impact its properties regarding its new mechanical specifications. Additionally, PLA is already soft at this temperature and can lead to blocking the tube just by piling up even without being completely melted.
It is safe to say that the hot end would perform correctly under these exacts specification and input. The behavior would quickly change as air temperature changes or heating temperature is increased.
It has to be noted that this geometry is not designed to be utilized without a fan and so it is not optimized for a passive cooling solution.
2. Hot End with fan
This configuration features a fan blowing air towards the heat sink, this helps taking the calories off by forcing air convection through the fins. This is how the actual design of the E3D is supposed to operate and ,before looking at the results, we can think that the efficiency of the heat sink will be much higher.
Indeed, we see on the left picture that the temperature on the heat sink is much lower than on the previous configuration. As the temperature does not exceed 40°C on the heat sink skin, we can safely say that the filament will not be melted inside the heat sink (this will be verified further in this article). The high temperatures are kept on the lower parts of the hot end, the heat break, the heat block and the nozzle. As the air is stopped from going up by the horizontal air flow of the fan, it cannot heat up the heat sink by convection, limiting the propagation of heat to only conduction through the materials.
The two pictures above show that the fan is very efficient in our application, indeed air temperature is high only really close to the heat block, above that part, the fresh air flow sweeps the ascending hot air just around the top of the heat break keeping the heat sink cool.
On the side picture, the temperatures of the two facing surfaces of the heat sink and heat block depict well the efficiency of the fan. There is almost a 200°C gradient between the two surfaces, clearly creating a known area where the phase changing of the PLA happens : the heat break.
Looking at the cut plot of the entire hot end, we can see that the PLA behaves a lot better with the fan activated. At the bottom of the heat sink, the PLA entering the heat break is at around 60°C and starts melting in the lower part of the heat break as expected. This way, when it is cooled down after a print job, the filament will not solidify in the heat sink preventing clogging when reheating.
We see that a fan is efficient enough for this application and at this kind of temperatures. A further study should be made to validate a worst case scenario behavior (250°C at heat block and 60°C air temp).
3. Hot End with fan and air duct
The final configuration features a fan and an air duct (which is often 3D Printed). We directly see, as expected, that the air flow is better channeled through the heat sink fins. However, the enhancement compared to the fan alone is not obvious at first sight. Indeed, the heat sink temperature seems very close to the previous configuration, and the overall temperature profile looks also similar.
Air around the heat block is disturbed by the fan air flow but this does not improve the efficiency of the system as the heat block does not really lose heat due to the fan air flow in the configuration without a duct.
Looking at the two facing surfaces of the heat block and the heat sink, we see that the temperature difference is again about the same as the previous configuration, with even a bit higher number for the lower heat sink surface as the air flow does not go through it directly.
The cut plot shows no real improvement compared to the second configuration, the PLA temperatures are about the same and even a bit higher. Again, there seems to be no gain from using a duct under these conditions.
This simulation shows the significance of the fan in this system. The efficiency and regularity of the air flow makes the fan compulsory. It allows a wide temperature range both of the hot end and the air. Indeed, printing some materials works better when in an enclosed space at a warm air temperature (usually 60°C), and this of course influences the hot end temperature and thus the PLA behavior. Also, printing other materials, like ABS or Nylon, requires a higher temperature of the heat block as their melting point is higher. Having a fan permits to work under all of these kind of circumstances.
However, as air ducts are widely used in the 3D printing world, it seems that their utility is not really proven, at least for this exact configuration. Further investigation will be made to assess the utility of the air ducts in different configurations.
Air flow illustration of the hot end with a fan
Air flow illustration of the hot end with a fan and an air duct
Air flow direction representation with temperatures
(@ 1m/s air speed, with and without air duct)