Can You Really Fry an Egg on a CPU?

Solving complex thermal models with computational fluid dynamics (CFD) requires a lot of processing power, and a central processing unit (CPU) under full load generates a fair amount of heat. But can you cook an egg on it? Search online and you can find videos of people attempting to cook on their processors—I wouldn’t recommend this as a cooling solution. However, just out of curiosity, I put this urban legend that you could cook an egg on the heat generated by a computer chip to the test using complex computational fluid dynamics (CFD) modeling with 3D computer-aided design. This article describes the model, simulations, and the ultimate conclusion.

Solving complex thermal models with CFD requires a lot of processing power and a CPU under full load generates a fair amount of heat. But can you cook an egg on it? Before you throw away your conventional heatsink and fan in favor of a multifunctional omelette, we’ll investigate what CFD in the FloTHERM XT tool can predict about the fate of your PC if you do so.

As a student intern at Mentor Graphics, I get the opportunity to create and play with interesting and unusual thermal models made in the software. For this concept, I started by creating a basic project comprising two cuboids and a cylinder to model a motherboard, processor, and egg. Then, I did some research—standard dimensions for motherboards, CPU thermal design power (TDP), and thermal data for eggs. Using this simple model, I was able to test things rapidly, get a rough feel for the range of temperatures and evaluate how feasible the concept was to model (Figure 1).

Figure 1. Model of an egg frying on a CPU in FloTHERM XT

I then created a more detailed model, starting by importing a board using FloEDA Bridge, a tool in FloTHERM XT that allows 3D CAD geometry to be produced from an intuitive 2D EDA environment. Using a topographical image of a representative mITX motherboard and a selection of known dimensions, I was able to accurately scale the image as a texture onto the board. This allowed the footprints of the board components to be positioned accurately against the image.

An excessively detailed board is not necessary for this model because we are only interested in the CPU temperature. To do this, the model needs to include the major sources of heat and any geometry that may alter the flow of air around critical components. The geometry around the CPU, such as the CPU socket, needs to be much more detailed because this has a significant effect on heat transfer by conduction. The CPU itself was modeled with accurate geometry and a two-resistor, network-assembly thermal model, based on an Intel Core i3-4130 as a representative mid-range desktop processor. Typical maximum Tjunc and Tcase are 90 °C and 72 °C [See reference 1].

The geometry far from the CPU, such as the I/O ports, can be modeled as simple cuboids that will present an obstacle to air flow on that edge of the board. Additional detail would just prolong the solver computation time with diminutive improvement in accuracy.

Limitations
One consideration is that the egg cannot quite be simplified to a material with thermal properties dependent on temperature, which FloTHERM XT is easily capable of modeling. For instance, apply heat to a block of aluminum and that energy will all eventually be transferred to the ambient air as it cools. Do the same to an egg and not all of the energy will be transferred to the ambient; cooking an egg is an endothermic process and a proportion of the energy will be converted to the change in chemical enthalpy of the egg as it cooks [see reference 2] (Figure 2).

So the question is, how significant is this lost “cooking energy” to our model? Well, after some research and calculations, not very. The specific enthalpy to denature the egg proteins during cooking is around 2.7 J/g for egg white and around 1.0 J/g egg yolk [see references 3, 4], so for an average 50 g egg, the total energy required is only around 80 J. This energy would contribute to a total drop in CPU temperature of approximately 6° C, but this removal of heat can only happen once per egg. Compare this to the 54 W TDP [reference 1] of the CPU under heavy load. After a few minutes of egg cooking, the heat dissipated is of the order of 10 kJ and the protein denaturation can only mitigate a negligible fraction of this.

The majority of the heat supplied ends up, well, heating the egg. The egg’s high water content gives it physical properties similar to water—slightly higher densities of 1130 kg/m³ for the yolk and 1133 kg/m3 for the white; specific heat capacities of 3.55-3.60 J/(kg K) for the yolk (increasing with temperature) and 2.55-2.75 J/(kg K) for the white; thermal conductivities of 0.550-0.558 W/(mK) for the yolk and 0.389-0.407 W/(mK) for the white (decreasing with temperature). The actual values used in the model were based on a moving average of reported experimental data [see reference 5].

Another factor not considered in the model is water in the egg heating up to steam, which rises, carrying away heat. But this is in small enough quantities compared to the flow of surrounding air that it can be ignored for now.

These two modeling limitations mean that the results are likely to be a slight over-estimation of the temperatures while the egg is still cooking, giving a worse-case scenario. This can often be beneficial; if the worse-case scenario is within thermal design constraints, the actual performance should be superior.

Above 65° C, the white begins to coagulate, and above 70° C, the yolk solidifies. Once solid and cooked, the egg is a much poorer heatsink; it is less thermally conductive, and convection within the egg no longer occurs. It is less dense due to water loss, and air/steam gaps are formed underneath the egg, insulating the CPU causing more heat to accumulate. Modeling the egg as always liquid gives a best-case scenario above 70° C.

Results
Unfortunately, the CPU junction temperature exceeds 90° C within 6 seconds, at which point the CPU clock would throttle down to reduce the thermal power and prevent damage to the system—less than ideal for a cooling solution (Figures 2 and 3). The egg would also burn and catch fire.

The central location of the CPU on the board and the large obstacles to air flow in the neighboring memory DIMMS and I/O ports mean limited cold air can passively flow over the hot egg by natural convection (Figure 4). Even adjusting the results for the modeling limitations described earlier, there is simply insufficient cooling.

Figure 2. Temperature surface plot of egg after 5 minutes of heating

Figure 3. Temperature against time for egg-cooled CPU

Figure 4. Particle plot, showing air flow around egg and CPU

For comparison, a new project configuration was created with the egg substituted for a standard Intel stock CPU cooler, radial heatsink, and fan (Figure 5). Once solved, this gave a junction temperature of 65.1° C after 5 minutes of the CPU being under full load at 54 W with the system approaching equilibrium (Figures 6–8).

Figure 5. Board with egg replaced with stock CPU cooler

Figure 6. Surface plot of stock-cooled CPU

Figure 7. Temperature against time for stock-cooled CPU


Figure 8. Particle plot showing air flow through stock cooler

Conclusion
The passive cooling of the egg cannot match the forced convection of the stock cooler. An egg-based cooling solution would only keep the CPU below the maximum 90° C Tjunc if the CPU performance were throttled down to 10-W TDP, so there are only possible applications in lower power environments with plenty of ventilation. With the requirement of frequently swapping out the egg, I can’t see this catching on. If the aim is to cook eggs though, CPUs certainly produce enough heat to do so; with thermal throttling, the processor acts as a thermostatically controlled surface at around 90° C, sufficient to cook on. If you value your computer, maybe consider buy a frying pan instead.

References
[1] Intel Corporation, 2014. ARK | Intel® Core™ i3-4130 Processor (3M Cache, 3.40 GHz) [online]. http://ark.intel.com/products/77480/Intel-Core-i3-4130-Processor-3M-Cache-3_40-GHz [accessed April 2015]
[2] J. McClements, University of Massachusetts, 2001. Thermal Analysis of Foods [online]. http://people.umass.edu/~mcclemen/581Thermal.html [accessed April 2015]
[3] C. Németh, K. Horváth, Á. Drobecz, L. Friedrich, K. Pásztor-Huszár, C. Balla. Calorimetric study of changes induced by preservatives in liquid egg products. Polish Journal of Food and Nutrition Sciences, 2010, Vol.60(4)
[4] A. Laca, B. Paredes, M. Díaz. Thermal behaviour of lyophilized egg yolk and egg yolk fraction. Journal Of Food Engineering, 2011 Jan, Vol.102(1)
[5] J. Coimbra, A. Gabas, L. Minim, E. Garcia-Rojas, V. Telis, J. Telis-Romero. Density, heat capacity and thermal conductivity of liquid egg products. Journal of Food Engineering, 2006, Vol.74