System Level Design and Analysis of the ACES Modular Ejection Seat Park O. Cover, Jr

Concurrent Technologies Corporation Abstract

McDonnell Douglas originally developed the Advanced Concept Ejection Seat (ACES II) in the 1970’s. The USAF has contracted Goodrich Universal Propulsion Company (UPCO) and Concurrent Technologies Corporation (CTC) to develop a modular version of ACES for the F-15. The modular design will allow seat removal without having to first remove the canopy from the aircraft.

ANSYSTM Workbench 9.0 shaped the modular design, enabling CTC to meet the demanding structural requirements for the new seat. Due to the complex loads this seat is subjected to, a system level analysis was developed. This linear analysis consists of approximately sixty structural components, uses a mixture of shell and solid elements and consists of approximately 320,000 nodes and 200,000 elements. Bonded contact was used to join the components together. This model allowed CTC to quickly and efficiently examine the effects of many load cases on the entire seat system when problem areas were identified and the geometry could be updated without erasing existing contact pairs, loads and boundary conditions. The ability to easily update and quickly solve the numerical model allowed for multiple iterations, resulting in an optimized design. The system analysis also allowed CTC to examine complex load cases, which were not possible during the original design in the 1970’s. Modifying and verifying multiple designs were easily accomplished using ANSYS Workbench. This approach ensures that the new modular design will meet all requirements during structural testing and qualification.

Introduction McDonnell Douglas originally developed the Advanced Concept Ejection Seat (ACES II) in the 1970’s. The USAF has contracted Goodrich Universal Propulsion Company (UPCO) and Concurrent Technologies Corporation (CTC) to develop a modular version of the ACES seat for the F-15. The system is designed to allow safe ejection from a variety of flight conditions such as zero velocity on the ground to Mach 1 high altitude flight [1]. The Modular ACES is designed to the same form, fit and function of the ACES II seat, using many of the same internal control systems. However there are many significant differences between the Modular seat and the ACES II seat. These changes were implemented in order to meet the following goals:

• Decrease maintenance time by configuring the seat to permit removal from its crew station without first removing a canopy, hatch or other overhead enclosure

• Improve maintainability of seat while still in the cockpit

• Increase safety by providing larger aircrew protection for through-the-canopy ejection

• Improve ergonomic accommodations (various height and weight aircrew)

The objective of this report is to present a portion of the work that was done to verify the structural integrity of the ACES Modular Ejection seat and to highlight the use of ANSYS Workbench 9.0 as a tool. This report will focus on the system level structural analysis that was performed using ANSYS Workbench.

Figure 1 shows the ACES Modular proof of concept mock-up. Figure 2 illustrates how the bucket can be removed from the seatback to facilitate removal of the ejection seat without having to slide the entire seat up the rails. After bucket removal, the seatback is removed by loosening several bolts that join the seatback with the rollers in the ejection rails. The areas around the modular joint are of particular interest because it is the main structural load path from the seat bucket to the seatback.

Figure 1. ACES Modular proof of concept mock-up.

Figure 2. Removal of seat bucket via the modular joint.

Background ANSYS Workbench shaped the modular design, enabling CTC to meet the demanding structural requirements for the new seat. Due to the complex loads this seat is subjected to, a system level analysis was developed. The loads are complex because they are applied to the entire seat. Without a system level analysis, it would be difficult to accurately determine the loads that are acting on each component in the structure. The system analysis that was developed was a linear analysis consisting of approximately sixty structural components. It used a mixture of shell and solid elements and has approximately 320,000 nodes and 200,000 elements. Bonded contact was used to join the components together. This model allowed CTC to quickly and efficiently examine the effects of many load cases on the entire seat system and to identify problem areas. The geometry could then be updated without erasing existing contact pairs, loads, and boundary conditions. The ability to easily update and quickly solve the numerical model resulted in an optimized design. The system analysis also allowed CTC to examine complex load cases, which were not possible during the original design in the 1970’s. Modifying and verifying multiple designs were easily accomplished using ANSYS Workbench. This approach ensures that the new modular design will meet all requirements during structural testing and qualification.

Analysis Description The approach to this analysis was to simplify the geometry so that a simulation of the entire seat could be performed in a relatively short time. A quick turn around time allowed for multiple design iterations to be examined.

Many of the modular seat components consist of thin sections, which allow for accurate representation via shell elements. Therefore, the first step was to create simplified surface geometry that was representative of the solid CAD geometry. For some very complex components, a mixed solid and surface model was created. This process involved converting the majority of the ACES modular structural components to surface models and assembling them using Pro-EngineerTM (ProE), then transferring this assembly to ANSYS Workbench. Geometry changes were implemented easily without having to redefine contact pairs, loads, boundary conditions or simulation commands. This was possible because the ProE surface models were parametric with the solid geometry, and because the ANSYS Workbench simulation was parametric with the ProE surface model.

Because of the way that shell properties are defined in ANSYS Workbench, care had to be taken in creating the surface models in ProE. For example, if a component had multiple areas with different thicknesses, they had to be defined as separate surfaces in ProE. When this multi-surface component is transferred to ANSYS Workbench, it appears as a sub-assembly allowing real properties to be individually assigned. In addition to this, components that are manifold, i.e. components that have a T-shaped surface intersection, had to be modeled using separate surfaces. Pure penalty bonded contact was used to join the surface edges together, creating a shell component that was representative of a solid component.

Figure 3 shows the geometry of the ACES Modular ejection seat ANSYS Workbench model. The simulation is composed of about 100 components, 60 of which are structural components while the remaining 40 were included to obtain the proper mass properties. Some of these components are broken down into sub-assemblies as discussed above, and because of this there are actually 186 geometric entities in the simulation.

Figure 3. ACES Modular ejection seat ANSYS Workbench geometry.

The parts within each sub-assembly were joined using edge to edge or edge to face automatic contact. Contact pairs between different components were manually created to ensure the connectivity of the structure was as accurate as possible. Most of the joints on the seat structure are made with rivets, so face-to-face contact was used to represent this type of connection. This is valid because a properly designed riveted joint will evenly distribute load over the areas being joined. This method also allows rivet patterns to be efficiently sized based on the contact reaction forces observed at each joint. The model has a total of 390 contact pairs, 202 of which join separate components together, while the other 188 join parts within sub-assemblies. When the geometry was updated these contact pairs remained in tact about 90 percent of the time. There were some instances where some existing contact pairs had to be recreated after a geometry update.

The model was meshed in ANSYS Workbench using aggressive shape checking and 49 mesh size controls which were applied to whole components and areas of interest. The resultant mesh was a mix between solid and shell elements, with about 200,000 elements and 320,000 nodes. Figure 4 shows the mesh of the ACES Modular ejection seat. This setup resulted in a run time of approximately 2.2 hours using a 3.6 GHz dual processor Intel / Windows workstation and the Sparse solver.

Figure 4. ACES Modular ejection seat ANSYS Workbench mesh.

The load cases that are presented in this report are ultimate load cases using equivalent static loads. The ultimate load is defined as the highest operational load that the system will experience, multiplied by a safety factor of 1.5. The margin of safety (M.S.) is calculated using the following formula:

1..max

−=SStuSM

Where Stu is ultimate tensile strength and Smax is the maximum calculated stress. The margin of safety must be greater than zero for the design to be acceptable.

Method Verification To verify that the simplified shell models were properly representing their solid counterparts, a test case was performed. The results of the test case showed that the calculated deflections and stresses of the solid component and simplified shell component were very similar. The models were constrained identically and loaded with a 100-lb force in the Z-direction. Figure 5 shows the Z-deflection of the solid component and the Z-deflection of its equivalent shell component. Figure 6 shows the equivalent stress for the same load conditions.

Figure 5. Comparison between solid component (top) and shell component (bottom), Z-deflection.

Figure 6. Comparison between solid component (top) and shell component (bottom), equivalent stress.

This test showed that the maximum Z-deflection for the shell model was 0.0017 in, and 0.0015 in for the solid model. This difference in stiffness was considered acceptable when weighed against the advantages of using the shell model. The equivalent stress contours for the two models are similar when plotted on the same scale, but the maximum stress is significantly higher in the shell model. The maximum stress occurs at the contact between the lower lug and the body of the component. This stress is caused due to the abrupt transition between the edges and surfaces that are in contact. Since the transitions are not this abrupt in reality, artificially high stresses at the contact edges are ignored when evaluating the system level analysis results. Sub-model analyses of high stress areas were performed using loads from the system analysis to verify that these areas satisfy structural requirements. Detailed descriptions of the sub-model analyses are beyond the scope of this document.

This report will focus on the final iteration of two specific load cases: catapult and forward crash. Many more load cases and geometry iterations were performed, but they are beyond the scope of this document. Goodrich UPCO provided general load data for each condition based on previous analysis and testing of the ACES II seat. The following sections describe how the loads and boundary conditions were applied to the ANSYS Workbench simulations in order to verify the structural integrity of the design.

Analysis Results & Discussion

Catapult Load When the ejection handle is pulled, a rocket propels the seat along the ejection rails and out of the aircraft. The maximum acceleration of the seat and occupant during this event is 24 g’s, which includes a safety factor. Figure 7 shows a schematic of the main forces acting on the seat system for this load case, i.e. the occupant weight (FR) and the rocket thrust (CKU). In this figure G is the acceleration of the system due to the rocket thrust.

Figure 7. Schematic of catapult load condition.

In order to accurately represent the response of the seat to this acceleration, the mass of the seat had to be correct. In addition to the mass of the seat structure with occupant, a large portion of the seat system mass comes from various sensors, control devices and other sub-systems. These sub-systems were represented in one of three ways; an ANSYS Workbench point mass, a block of solid geometry with appropriate mass (which will be referred to as a lump mass from this point forward) or a force equivalent to the component mass multiplied by the acceleration. Using these methods allowed more flexibility when setting up the boundary conditions for the model. For example, if a sub-system was mounted to two or more brackets, a point mass could not be used, since it cannot be applied to multiple components.

For the catapult load case there were six point masses, six lump masses, six equivalent forces and a global acceleration of 24 g’s. Of the six support rollers on the seatback, only the lower four were used to constrain the model in the X and Y direction. The bottom of the rocket was constrained in the Z direction. Figure 8 shows the ANSYS Workbench Environment view of the catapult loads and boundary conditions.

Z X

G

FRCKU

Figure 8. Catapult loads and boundary conditions.

Catapult Results Several geometry iterations were examined before converging on a successful solution. The equivalent stress results from one of the first iterations are shown in Figure 9. The areas that are purple are above the allowable stress of 72 ksi, the ultimate tensile strength for high strength aluminum [2]. The deflection scaling for all of the results plots are greater than one to exaggerate seat deformation. Based on the contours that are shown in Figure 9, it is clear that there are several areas that do not meet the structural requirements. This information was used to develop a design that successfully meets the requirements.

Z X

Z Acceleration

Figure 9. Equivalent stress, initial seat design, catapult load case. The results of the final iteration of the catapult analysis show that the ACES Modular ejection seat satisfies the structural requirements with a positive margin of safety. Figure 10 shows the equivalent stress of the entire seat with the maximum contour set to 72 ksi. As this figure shows, the maximum equivalent stress is greater than the allowable stress of 72 ksi, however this occurs at an abrupt transition area. As described earlier, these unrealistic stresses are ignored for the system analysis. More importantly, there are no large areas that exceed the maximum allowable stress, as were seen in the earlier seat design.

Views of each component were set up in the results section of ANSYS Workbench so that each component could be quickly examined in detail. Figures 11 and 12 show detail views of the modular joint area. Using these views, the margins of safety were calculated for each structural component. Because the reported maximum typically occurs at an abrupt transition area, engineering judgment was used to locate the true maximum stress of each component. For the parts in Figures 11 and 12, the maximum stress was found to be 69.1 ksi and 62.4 ksi, respectively. These stress values correlate to margins of safety of 0.04 and 0.15, respectively. All other structural components were examined in this fashion and found to be acceptable.

Figure 10. Equivalent stress, overall seat, catapult load.

Figure 11. Equivalent stress, modular seatback connections, catapult.

Artificial Contact Stress

Max stress occurs at hole

Figure 12. Equivalent stress, modular bucket sides.

Artificial Contact Stress

Forward Crash Load To ensure the safety of the aircrew during extreme conditions, the ejection seat structure is required to withstand a forward crash equivalent to a static 40-g ultimate load. The restraint system for the seat occupant must also withstand this high load. Figure 13 shows a schematic of the main forces acting on the seat system for this load case, i.e. the occupant weight and g-load. In this figure G is the acceleration of the system, FS is the occupant reaction on the shoulder harness and FL is the occupant reaction on the lap belt.

The mass of seat components was represented in a fashion similar to the catapult load case. Due to the differences in the load cases, this analysis used five point masses, six lump masses, 13 equivalent forces and a global acceleration of 40 g’s. All six support rollers on the seatback were used to constrain the model in the X and Y direction. The bottom of the rocket was constrained in the Z direction. Figure 14 shows the ANSYS Workbench Environment view of the forward crash loads and boundary conditions.

Figure 13. Schematic of forward crash load condition.

Z X

G

FL

FS

Figure 14. Forward Crash loads and boundary conditions.

Forward Crash Results The same geometry iterations that were examined for the catapult load condition were also examined for the forward crash condition. The equivalent stress results from the initial geometry iteration are shown in Figure 15. The deflection scaling for all of the results plots are greater than one to exaggerate seat deformation. Based on the contours that are shown in Figure 15, it is clear that this geometry iteration does not meet forward crash requirements or catapult requirements. After all the load cases were examined to determine the weak areas of the design, redesign efforts focused on fixing these issues. The headrest area and the upper seatback area required the most attention based on the results of all these initial runs.

Forward Deceleration

Z X

Figure 15. Equivalent stress, initial seat design, forward crash case.

The results of the final iteration of the forward crash analysis and further sub-analyses show that the ACES Modular ejection seat satisfies the structural requirements with a positive margin of safety. Figure 16 shows the equivalent stress of the entire seat with the maximum contour set to 72 ksi. Again the maximum stress is greater than 72 ksi, but as described earlier these unrealistic stresses are ignored for the system analysis since they occur at an abrupt transition area.

Figure 17 shows a detail view of the modular joint area. The joint is loaded primarily by the lap belt in this load case, and has several small areas which exceed the allowable stress of 72 ksi. Figure 18 shows a detail view of the headrest support structure, which must support the shoulder harness loads. Due to its geometry, it could not be simplified to a shell model. Figure 19 shows the headrest slide structure, which was simplified to a shell model, but lost a significant amount of fidelity in this conversion.

Based on the results of this load case, it was determined that sub-analyses were required to further study the modular joint and headrest areas. Contact reaction forces from this system analysis were used to accurately represent the loading conditions on each sub-section. The results of these sub-analyses showed that the modular joint and the headrest structure successfully satisfy the requirements with positive margins of safety. Further details of these sub-analyses are beyond the scope of this paper.

Figure 16. Equivalent stress, overall seat, forward crash load.

Figure 17. Equivalent stress, modular seatback connections, forward crash.

Figure 18. Equivalent stress, headrest support structure, forward crash.

Figure 19. Equivalent stress, headrest slide structure, forward crash.

Conclusion This work focused on the development of a system level analysis of the ACES Modular ejection seat. The outcome of this effort showed that a very complex analysis model could be built and solved relatively quickly, enabling an optimized ejection seat design. In addition, this method highlighted critical structural areas of the design, which were later analyzed using sub-analyses. This work resulted in a structure that meets the program goals of decreasing maintenance time, improving maintainability, increasing safety and enhancing the ergonomics of the seat.

References 1) “Advanced Concept Ejection Seat (ACES-2) Loads and Stress Analysis”, McDonnell Douglas

Corporation, 1972.

2) “Metallic Materials Properties Development and Standardization”, Office of Aviation Research, Washington, D.C. 20591.