The use of unmanned aerial vehicles (UAVs) has been increasing around the world as many countries have been rapidly adding UAVs to their arsenals during the past decade. One of the most critical challenges facing UAV designers is icing. Icing has long been known to be dangerous for aircraft. A pilot, though, can recognize icing conditions and compensate for them by either activating de-icing systems or by adjusting to the changing flight characteristics of the aircraft and removing the aircraft from the icing conditions. The current generation of US Military UAVs deployed in Afghanistan had issues with icing that caused many crashes, costing millions of dollars. This led to the development of UAV design variants with deicing systems.

Computational Flow Dynamics

Figure 1. Example of typical CAD geometry for a Predator-like UAV.
Typically the amount of time it takes to predict the thermal and airflow performance of an aircraft using a traditional CFD (Computational Fluid Dynamics) code is daunting. A significant amount of time is spent on creating an accurate mesh to create a useful model. Achieving this mesh is especially difficult for the curved surfaces prevalent on aircraft but a sufficient mesh in the boundary layer is critical to the accuracy of the result. This type of analysis isn’t commonly used to aid in the design process but rather to validate the final design.

A design engineer, however, needs to be able to simulate a design in all operating conditions and configurations to gain the most insight. CAD-embedded CFD allows fast, accurate CFD simulations on existing CAD geometry, and knowledge of the numerical methods used in complex CFD analysis isn’t necessary.

Imagine the scenario that likely played out when these UAVs started having icing problems. First, senior management was notified who passed the issue over to the engineering managers, who would then bring together an engineering team to design a remedial solution. The engineers would be working to understand the issue as well as many different de-icing systems, each with its own pros and cons.

Because UAVs are surveillance aircraft designed to remain in the air for long periods, any de-icing system must be as light as possible to maintain fuel efficiency. At this point, CAD-embedded CFD can save the UAV manufacturer a lot of money. By simulating the different design options at an early stage, the engineers can eliminate the wrong design paths early before a significant amount of engineering time has been spent working on dead-end designs and it can more quickly narrow the engineering focus onto the designs that will work.

Figure 2. CAD-embedded CFD analysis shows that the older model de-icing system is inefficient.
As an engineer tasked with simulating the possible de-icing systems, one of the biggest hurdles would be creating the model. With CAD-embedded CFD, this task is greatly reduced because the UAV has already been designed (Figure 1—this actual generic geometry was obtained from,  because CAD geometry exists for everything. The only geometry that needs to be created is for the de-icing system. When CFD software resides inside a CAD package the design engineer has already spent thousands of hours using, this task won’t take very long because the engineer is using the tool he or she knows best.

Sample Systems

Let’s look at three examples of how CAD-embedded CFD would help in analysis of three different de-icing systems.

The first de-icing system involves placing electrical heating elements in the leading edge of the wing. These systems have been around since the 1940s, and they operate by heating up the surface above the melting point of the ice. Modeling this with the CADembedded CFD software was very simple. All that was needed was to split the surface to create the leading-edge zone where heat would be applied.

The CAD-embedded CFD analysis showed that while the surface where the heating was applied was kept above the freezing temperature, behind the leading edge of the wing was a different story. The original assumption was that the heating element would cause a heated boundary layer that would protect the rest of the wing. Further CFD simulations showed that to achieve that, the front surface would have to be >100 °F and would require ~2,700W of energy to have warm air cover the 2.4 ft. wing section used for the simulation (Figure 2).

Figure 3. CAD-embedded CFD simulation of the ThermaWing™–like zoned de-icing system shows that it should perform as promised by the manufacturer [*].
That is not a good option because a higher energy requirement means more fuel would be needed. The problem is that while the leading edge is kept too hot for ice to form, water is forming on the leading edge. Because of the airspeed, this water will run back to the unprotected surface of the wing and freeze. If we are going to design a deicing/ anti-icing system, we don’t want to design one with such an obvious flaw.

Another similar concept for a de-icing system was developed by design engineers at Kelly Aerospace called the ThermaWing™ [*]. The elements of the system are basically the same. It uses electrical heating elements on the surface, except that they are separated into two zones. Right on the very front of the wing is the always-warm zone. Behind this, a secondary heating zone is created by cycling heating on and off. Once a layer of ice has formed, the heater is cycled on. The secondary zone heater doesn’t have to melt all the ice; it just melts a thin amount of ice at the surface. Once a thin film of water has been created between the surface and the rest of the ice, the bond between the ice and wing is broken and the ice will be shed. Then the cycle repeats. This system seems like it would be more energy efficient because it isn’t on continuously.

For analyzing this design with the CADembedded CFD software, the wing surface was split into three zones: the alwaysheated zone, the periodically heated zone, and the aft unprotected zone. A block of ice also had to be modeled on top of the wing surface, which was assumed to be 5mm thick. All these geometry changes were quick and easy to make in the CAD system. The analysis was set up to be a transient simulation, with the secondary heater turned on at t = 0.

The results showed that this system would work and requires less energy than the previous system at ~1,200 W when both heating zones are activated for the test section. Since the majority of that energy is used periodically for the secondary heating area, the time averaged energy use would be much lower, but the cycle rate would depend on the ice accumulation rate so averaged power dissipation was not calculated. The CADembedded CFD simulation showed the average temperature of the ice-to-wing surface interface reaches its melting temperature at about 1 second, the exact amount of time claimed by Kelly Aerospace in their literature. Figure 3 shows the temperature results.

Figure 4. CAD-embedded CFD analysis shows that the heated air tube system causes the bottom of the wing to be hotter than the top.
If this design was chosen as the one to proceed with, further CAD-embedded CFD analysis could find the optimized areas for the heating zones, the results at different altitudes, airspeeds, angle of attack, and the total wattage the system would need to protect the aircraft.

A third de-icing design analyzed here is a heated air tube system, in which pressurized hot air is sent through a tube that runs under the surfaces that are to be heated. The tube has holes to spray the hot air onto the surfaces. Typically, this hot air is provided by air bleed from the gas turbine engine. The UAV in this case has a piston engine, which makes providing the bleed air more complex. But the avionics that make a UAV possible provide an endless amount of hot air. Using waste heat would mean that no extra fuel would be burned to solve this design issue.

This particular CAD model didn’t have the internal details of the UAV wing structure, which would not be the case for an engineer working at the UAV manufacturing company. For this design, the wing body was shelled out and a hollow pipe was created to follow the sweep and dihedral of the wing. The initial position of the hot air tube and the position of the tube holes were arbitrarily set because the design was just at the concept stage. If this design was chosen to move past the concept stage, further CFD simulations could easily be run to optimize the position of the tube, hole positions and diameters, and tube pressure and hot air-flow rate requirements to protect the entire UAV.

The results show that the surface isn’t a uniform temperature. The bottom of the wing is hotter than the top, and there are clear hot spots created by the hot air jets. The surface is hotter than it needs to be, and the flow rate or hot air temperature could be reduced. This system transferred ~780 W to the 3.3 ft. wing section used for this simulation (Figure 4). The downside of this design was that the rear portion of the wing was left unprotected.


These three examples clearly illustrate that CAD-embedded CFD software such as Mentor Graphics’ FloEFD can help significantly in solving engineer design problems early in the design stage with turnaround times of several hours before investment is made in manufacturing. Simulating different design options and choosing which one is the best course to follow is a big paradigm shift for most defense companies where designs are simulated by analysis groups near the end of the design cycle. CFD is therefore now much more accessible and easy to use for a designer because the CFD analysis software is automated and embedded in already familiar CAD systems.

This article was written by Travis Mikjaniec, Application Engineer, Mentor Graphics Mechanical Analysis Division (Wilsonville, OR). For more information, Click Here .


* Kelly Aerospace Thermal Systems,