System Analyzer for a Bioinspired Mars Flight Vehicle System for Varying Mission Contexts

Abstract

The Marsbee is a novel bioinspired flapping flight vehicle concept for aerial Mars exploration. The Marsbee design addresses the challenges of flying on Mars by mimicking the unsteady lift generation mechanisms seen in terrestrial insects To enable the comparison of the Marsbee system to other flying Mars exploration concepts, a study was performed that employs a Multidisciplinary Design Optimization architecture to analyze and optimize the Marsbee system to suit a wide variety of missions. This study developed an analyzer for a Multidisciplinary Design Feasible (MDF) architecture, as well as explored the design space and attributes necessary in an objective function for Mars flying system missions. The analyzer is based on physical models developed in previous studies. Its functionality was demonstrated by analyzing 100,000 randomly generated designs, with design variables close to a prototype Marsbee tested in Martian density conditions. These results show that by using flexible wings rather than rigid wings the maximum flight times increased from 53 minutes to 114 minutes, and the maximum payload masses increased from 28 grams to 61 grams. These are competing effects and cannot be maximized simultaneously. The results of this study will be used to determine the optimal Marsbee system.

I. Introduction

Mars exploration has gained increasing attention from both the public and private sectors. Rovers, satellites, and most recently the InSight Lander, have been sent to Mars on exploratory missions [1]. The upcoming NASA Mars 2020 mission includes a rover that will carry the Mars Helicopter Scout, a solar-powered rotorcraft that is set to be the first vehicle to demonstrate aerodynamic flight on Mars [2].

Flight on Mars would open up a new dimension of reconnaissance and enhance the overall performance of a rover mission. Currently, rovers have limited active periods of about 4 hours every sol (Martian day) when the Sun is at its peak. Scientists and engineers on Earth uplink defined tasks at sunrise [3]. The rover performs the commanded tasks during the 4-hour period while carefully avoiding obstacles using instruments such as the Hazcams. The data gathered during the day is downlinked to Earth at sunset. Communication with Earth resumes at sunrise when the new tasks are uplinked to the rover. Communication is one of the major challenges in driving a rover from Earth due to the one-way time delay of between 2 and 20 minutes between the two planets [4]. Both communication delay and obstacle avoidance slow down the rover’s exploration. The average distance traveled by the Curiosity rover per sol is just 32.1 m [5], lower than the maximum of 200 m per sol the drive system was designed to operate at [6].

One of the main challenges rovers face is incomplete knowledge of the terrain, which increases the risk of encountering situations that could damage the rover. The forward sensing abilities of a rover is especially limited in rocky or hilly terrain. Satellite images of Mars are also limited by current technology. The Mars Reconnaissance Orbiter provides the best satellite view of the Martian surface at 0.5 m/pixel [7, 8]. A flying scout vehicle would provide more detailed information on the localized surroundings of the rover including “over the hill” awareness. The flier could capture high resolution surface images on the order of mm/pixel using current CMOS imagining sensor technology [9]. Thus, a flying scout vehicle would potentially increase the speed at which a rover could explore the Martian surface, thereby increasing the science return per unit time of the entire mission.

Flight on Mars is a highly interdisciplinary challenge. In addition to the traditional engineering challenges, the flying system must interact with both the environment and the rover in a highly coupled system of systems. To aid in the system design process, an understanding of the flying system as well as the desires of the stakeholders is required. In this research, an analysis of a flying system is developed, the design space is explored, and attributes critical in the objective function are discussed.

II. Background

NASA has four primary goals regarding Mars exploration: life, climate, geology, and humans [10]. Types of missions pertaining to the first goal of life are searching for life sources (e.g. water and chemical energy) and searching for life signs (e.g. carbon and bio-signatures of current or past life). Types of missions pertaining to the second goal of climate are studying the dust storms to understand how they develop and grow. Types of missions pertaining to the third goal of geology are studying the Martian rocks and the planet’s magnetic field. A type of mission pertaining to the fourth goal of humans is studying the radiation environment on Mars. A suite of systems have been employed by NASA through various missions to accomplish these goals.

Martian exploration has been primarily conducted via rovers and orbiters. These vehicles have gathered and returned data to scientists on Earth since the 1960s [11]. Flying systems would offer additional capabilities to rovers and orbiters but have their own challenges. While the Martian gravity is one-third of Earth’s [12], the Martian surface atmospheric density is only 1.3% of Earth’s sea level density [13]. Designing traditional fixed wing aircraft for Mars is also challenging because of the low Reynolds environment [14–18]. Several research projects have been conducted in the past decades exploring the feasibility of flight on Mars, yet no vehicle has ultimately been fabricated and flown on Mars [2, 19–22]. However, flight on Mars could be realizable with bioinspired flapping flight vehicles, which are known to operate in low Reynolds number environments on Earth [6, 8–10, 15].

This paper focuses on the Marsbee, a novel bioinspired flight vehicle concept meant to aid a rover in the exploration of Mars. Figure 1 [24] illustrates how the Marsbee system would enhance the surveillance capabilities of a rover. Due to its many modes of flight (e.g. hover, forward flight, climbing, and gliding), the Marsbee would be capable of performing a variety of different scouting missions. The Marsbee is meant to fly in the low Reynolds number Martian atmosphere by capitalizing on the unsteady, high-lift generating mechanisms utilized by insects, such as wake capture, delayed stall, and wing-tip vortices [6, 9, 15]. To generate these same beneficial aerodynamic forces, the Marsbee will retain dynamic similarity to terrestrial fliers, preserving all the pertinent non-dimensional parameters such as wing-tip Mach Number [17]. Using high-fidelity Navier-Stokes simulations, previous Marsbee studies have demonstrated that the average lift produced by the wings was sufficient to support the weight of the vehicle under Martian atmospheric conditions [8–10]. A prototype flapper was constructed and the aerodynamic forces generated in Martian density conditions were measured [24, 25]. The results of these proof of concept experiments confirmed the results of the Navier-Stokes simulations and demonstrated that the mean lift of the flapper was sufficient to counterbalance the vehicle weight on Mars. This supports the feasibility of the concept.

Fig. 1.

An artisťs’ rendition of the Marsbee vehicle concept.

The analyzer developed in this paper uses several physical models coupled together, with the aerodynamics model being the most complex. Flapping aerodynamics models used in previous studies were adapted to fit the current work. The analyzer uses three increasing fidelities of aerodynamics models: the zeroth-order model (0th), the quasi-steady model (QS), and the Navier-Stokes model (NS). The 0th order model [17] leverages the definitions and scales up the bounds of non-dimensional numbers (i.e. Reynolds number, reduced frequency, lift coefficient) relevant to insect flight to the Mars flight regime. This is a simple analytical model that only calculates mean aerodynamic lift and does not calculate drag or aerodynamic power. The QS model represents the next step up in fidelity and is based on the work of Lee et al. [26]. This analytic model calculates the lift and drag of the wing by considering translational, rotational, and added mass components of quasi-steady, insect-scale aerodynamics. This model has been shown to agree reasonably well with low Reynolds number insect flight on Earth, but it does not account for the wake capture lift enhancement mechanism or the nonlinear wing-wake interaction [6, 15]. The highest fidelity aerodynamics model available is a well validated Navier-Stokes equation model [8–10]. This NS model, which can be run in either 2D or 3D, and can capture all of the unsteady, low Reynolds number aerodynamic mechanisms found in insect-scale flapping wing flight. It has also shown good agreement with 3D experiments and simulations.

Another of the physical models used in the analyzer is the inertial power model. Using the wing kinematics and mass moments of inertia, this model calculates the inertial power required to move the wings about the two axes of rotation and thus has two components: flapping inertial power and pitching inertial power. In terms of vehicle design, the inertial power for Martian fliers is a larger contributor to the total power for flapping wings than Earth fliers due to the lower atmospheric density (thus lower aerodynamic force) and larger wings required to produce equivalent lift (thus more mass to move) [16]. Wing flexibility and energy capturing devices (e.g. integrated springs) can greatly reduce the inertial power required. This study considers four different wing designs, each having a different inertial power: rigid, flapping only, pitching only, and flexible. These different designs are summarized in Table 1. The rigid wing represents the worst-case inertial power scenario, with no energy capturing mechanisms utilized. The flapping only wing is able to passively pitch (accomplished by terrestrial fliers with spanwise flexibility [14]), thereby removing any pitching inertial power. The pitching only wing is able to passively flap (usually accomplished by a combination of chordwise flexibility and a power-capturing device such as a spring at the root of the wing), thereby removing any flapping inertial power needed. The flexible wing represents the best-case inertial power scenario, with zero inertial power required to move the wing – both flapping energy and pitching energy are stored and reused.

Table 1.

Wing designs used in this study and their inertial power components.

	Flapping Inertial Power	Pitching Inertial Power
Rigid	Yes	Yes
Flapping Only	Yes	No
Pitching Only	No	Yes
Flexible	No	No

The Marsbee project is a multidisciplinary effort that involves several branches of engineering - mechanical, aerospace, and systems - working in tandem. Multidisciplinary Design Optimization (MDO) [27] provides a methodology for ensuring consistent couplings and the identification of the optimal design. MDO traditionally works in a design space that is bound by constraints. The design space is formed from an objective function created by the designers before employing MDO, and then constraints are imposed on the design space [28]. Several MDO architectures exist [21, 22], but the architecture used in this current work is Multidisciplinary Design Feasible (MDF), which was categorized by Yi as a single-level architecture to achieve the best design compared to two other architectures [29]. MDF was chosen because it produces a design that is consistent with the design variables at each iteration of the optimization [27].

III. Methodology

The system analyzer is incorporated into an MDO framework to enable optimization of the Marsbee across a wide variety of mission contexts. The analyzer of this study was designed to account for different modes of flight, including hover, forward flight, and climb. The Marsbee vehicle can accomplish a variety of scouting missions by employing one or more of these flight modes. However, for this discussion we chose to focus on hovering flight only.

The first step in developing the system analyzer was to establish the behavior variables. The behavior variables are performance metrics that could apply to any flying vehicle. Next, the physical models used inside the analyzer were chosen based on the desired behavior variables. As discussed in the background, this study leverages physical models developed in previous work. The set of inputs into the analyzer were derived from the inter-model interactions and flow of information and coupling variables between these physical models. These inputs can be divided into two categories: environmental parameters and design variables. Environmental parameters are defined as parameters that are not under the control of the designers, such as atmospheric density. Design variables are the vehicle properties that the designers can directly change. The set of all possible design variables and their ranges make up the design space of the vehicle. Some assumptions were made to reduce the design space.

After the analyzer was fully developed and all the internal models were individually confirmed to be working properly, the functionality of the full analyzer was tested. A reasonable demonstration subset for all the designs variables was chosen based on the Marsbee flapper tested in previous vacuum chamber experiments [24, 25]. This small subset of the entire design space still represented 100,000 randomly generated designs. These designs were fed into the analyzer to produce their behavior variable values. Based on the resultant behavior variables, designs that violated model assumptions or feasibility constraints were identified. The solution space was then explored.

IV. Results and Discussion

Determine Behavior Variables

The first step in defining the analyzer was to choose the desired behavior variables of the vehicle. In this study the desired behavior variables, shown at the bottom of Table 2, are the payload mass and flight time of the vehicle. These behavior variables were chosen because they measure the general aerodynamic performance and can be used to compare different flying vehicles regardless of the specific mode of flight employed. They will be discussed in relation to potential objective functions later in the section.

Table 2.

Environmental parameters, design variables, and behavior variables of the analyzer.

	Category	Variable	Units	Min	Max
	Environmental Parameters	Atmospheric Density	kg/m3	0.0142
		Atmospheric Viscosity	kg/(m·s)	1.5e-5
		Speed of Sound	m/s	244
		Gravitational Acceleration	m/s2	3.72
Design Variables	Wing Morphology	Span	m	0	∞
		Mean Chord	m	0	∞
		Wing Thickness	m	0	∞
		Wing Density	kg/m3	0	∞
	Wing Kinematics	Flapping Frequency	Hz	0	∞
		Flapping Amplitude	°	0	90
		Pitching Shape Factor	--	0	∞
		Pitching Amplitude	°	0	90
		Pitching Phase Offset	°	−180	180
	Power Generation	Battery Specific Energy	Wh/kg	0	∞
		Depth of Discharge	--	0	1.0
	Other Variables	Mass of Battery	kg	0	∞
		Additional Mass Required to Fly	kg	0	∞
		Additional Power Required to Fly	W	0	∞
	Behavior Variables	Payload Capacity	kg	--	--
		Flight Time	s	--	--

Determine Assumptions

The analyzer was designed to be adaptable to different modes of flight, but for simplicity the hovering flight mode was chosen for explaining the analyzer. Furthermore, the analyzer is capable of using a variety of fidelities of aerodynamic models, but only the QS model is discussed here, as it is the lowest fidelity able to output aerodynamic power (a parameter important for calculating the behavior variables). The wings were assumed to be two, symmetric, quarter-elliptical panels with a constant thickness and a homogenous composition. The flapping and pitching kinematics were both symmetric, with flapping following a purely sinusoidal motion and pitching following a modified sinusoid (sine to square wave). The deviation motion was assumed to be zero. A battery was assumed as the power source. The current battery model ignores complex effects such as the relationship between specific power, specific energy, and operational temperature. The fluid flow was considered to be incompressible (justified as long as the wing tip Mach number remains below 0.3), and the coupled wing-body dynamics were ignored.

Determine Inputs

The environmental parameters are listed at the top of Table 2. Atmospheric wind gusts were ignored in this study. The design variables are summarized in the middle of Table 2. The ranges listed in Table 2 represent the physical lower and upper bounds placed on these inputs. The design variables are sorted into four categories: wing morphology, wing kinematics, power generation, and other variables. Many of design variables maintain their aerodynamic community definitions, however a few require clarification. The pitching shape factor transforms the pitching motion from a sinusoid (at 0) to a square wave (at ∞). Specific energy is the measure of total usable energy per unit mass stored in a battery. It is heavily influenced by the battery chemistry, specific power, and operational temperature – all of which are purposely left out of the current model for simplification. Depth of discharge refers to the maximum percent of total energy that can be drained from a battery before it needs to be recharged. Batteries that are consistently drained below this amount have a reduced lifespan. Aside from the wings and the battery, there are other components required for flight (e.g. actuators, frame structures, guidance & navigation equipment, and communication systems). Each of these add mass and some require power. Since the specifics of these additional components are not yet known at this stage, they are estimated from existing flapping flight vehicles and cumulatively represented by the additional mass and additional power required to fly terms.

Create System Analyzer

After establishing the inputs and the outputs of the analyzer, the physical models that constitute the analyzer were defined, and the coupling and information flow between these models were determined. Figure 2 is a flow chart of the couplings. Note that the color-coding of this figure matches that of Table 2. The analyzer consists of six model groups, each of which are based on physical equations: the wing models, kinematics model, aerodynamic model, payload model, power models, and battery model. The wing models calculate parameters used in other models like wing tip Mach number and moments of area. The kinematics model constructs time histories of the angular position, velocity, and acceleration of the wings. The aerodynamics QS model [26] outputs lift and aerodynamic power. The payload model takes the difference between the lift from the aerodynamic model and the total mass required for flight. Any excess lift can be used to carry additional mass as a payload. The power model takes the four different wing designs introduced in the background (see Table 1) and from these calculates four inertial powers, four total powers, and four average powers. Using these average powers, along with the battery design variables, the battery model is used to predict the total flight time of the vehicle. There will be four predicted flight times corresponding to the four different wing designs.

Fig. 2.

Analyzer Couplings.

Assess System Analyzer

After assembling the models together, the analyzer was used to explore the design space. For this exploration, a vehicle design randomizer was implemented to determine random values of the design variables. Table 3 shows the ranges for design variables used in this assessment. The ranges for design variables were about ±20% of the values from the prototype Marsbee flapper tested in a vacuum chamber at Martian pressure conditions [25]. Vehicle designs were generated by randomly selecting values in these ranges (with an equal probability distribution). The environmental parameters were fixed. A total of 100,000 designs were generated. Each design was then fed into the analyzer, which output the solution for flight time and payload mass.

Table 3.

Ranges of design variables used for the assessment of the analyzer.

	Category	Variable	Units	Min	Max
	Environmental Parameters	Atmospheric Density	kg/m3	0.0142
		Atmospheric Viscosity	kg/(m·s)	1.5e-5
		Speed of Sound	m/s	244
		Gravitational Acceleration	m/s2	3.72
Design Variables	Wing Morphology	Span	m	0.1	0.2
		Mean Chord	m	0.05	0.1
		Wing Thickness	m	20e-6	30e-6
		Wing Density	kg/m3	1400	2100
	Wing Kinematics	Flapping Frequency	Hz	0	30
		Flapping Amplitude	°	50	80
		Pitching Shape Factor	--	0	5
		Pitching Amplitude	°	35	55
		Pitching Phase Offset	°	−10	10
	Power Generation	Battery Specific Energy	Wh/kg	120	180
		Depth of Discharge	--	0.6	0.7
	Other Variables	Mass of Battery	kg	1.3e-3	2e-3
		Additional Mass Required to Fly	kg	0.0027	0.0027
		Additional Power Required to Fly	W	0	0

The designs were then deemed acceptable if the model constraints and feasibility constraints of the solution were not violated. The model constraints were the non-dimensional constraints assumed in the aerodynamics model: Reynolds number between 10 and 10⁴ (the insect flight regime) and Mach number less than 0.3 (incompressible flow). The feasibility constraints were the minimum desired performance requirements of the vehicle: payload mass above 1 gram and flight time above 1 minute (recall that each design has four flight times based on wing design, as summarized in Table 1). The unacceptable designs were separated from the acceptable designs using a two-step elimination process: first, the model constraints were checked, then the feasibility constraints were checked. Table 4 shows how many designs were eliminated at each step and how many designs remained for each wing type. Note that each row sums to 100,000 – the total number of randomly generated designs. The number of designs rejected on the basis of violating the model constraints does not change based on the wing type. This is because the non-dimensional number constraints are independent of the inertial power required to move the wing. In analyzing this further, it was found that the only model constraint that was violated by any of the designs was that of the lower bound to Reynolds number. Out of all 100,000 designs, the maximum Mach number was 0.211 and the maximum Reynolds number was 2328. Further, only about 1% of the designs were rejected for violating model constraints, whereas between 62% and 77% of the designs were rejected for violating the feasibility constraints.

Table 4.

Number of Acceptable and Unacceptable Designs

	Acceptable Designs	Designs that Violate Feasibility Constraints	Designs that Violate Model Constraints
Rigid Wing	22352	76684	964
Flapping Only Wing	32337	66699	964
Pitching Only Wing	25774	73262	964
Flexible Wing	37305	61731	964

Figure 3 shows the distribution of the two behavior variables, with each point representing a single randomized design. There are four subplots in this figure – one for each wing type. The wing model color-coding matches that used in Table 4. Along with the acceptable solutions, the solutions that violate either feasibility or model constraints are also indicated. Each of the data series have well-defined borders between the regions of feasibility violations and model violations. This change correlates with Reynolds numbers dropping below the cutoff of 10, which is associated with the wings moving very slowly. There is also a clear distinction between the acceptable designs and the designs that violate the feasibility cutoffs. Moreover, we can deduce which of the cutoffs have been violated. The data to the left of the vertical border of the good designs violates the flight time cutoff whereas the data below the horizontal border of the good designs violates the payload cutoff. The flexible wing (magenta data series) only violates the payload mass cutoff, implying that it is the design space itself that is keeping the flight time from going below 1 minute. Note that the flight time axis is on a logarithmic scale. The flight times for the unacceptable designs become unphysically large. This can be attributed to the fact that payload mass dips into the negative at these large flight times, which is clearly a non-physical result. Interestingly, the only time the model assumptions are violated are when the payload times are negative, which would also violate the feasibility constraints.

Fig. 3.

Good and Bad Designs for Different Wings.

Figure 4 shows the performance plots of the different wing types, with each point representing an acceptable design solution. The points representing designs that violate the imposed constraints have been eliminated. Both axes are linear and the color-coding matches that of Table 4 and Fig. 3. For each of the wing types, we can see a Pareto frontier forming one of the boundaries of the solution space. The other boundaries are due to the elimination criteria discussed above.

Fig. 4.

Performance of Different Wings.

As expected, we see that there is a tradeoff between payload mass and flight time: as payload mass increases, flight time decreases and vice-versa. We can also compare the impact on performance the different wing types have. Tables 5 and 6 summarize the performance values seen in Fig. 4 for flight time and payload mass, respectively. The averages in these tables are derived from all of the data points for a wing type. As we go from the rigid wing (the maximum inertial power scenario) to the flexible wing (the minimum inertial power scenario), flight time sees a magnitude increase, from an average of 2.68 minutes for the rigid wing to 26.11 minutes for the flexible wing. Comparing the maximum flight times, we see more than a twofold increase from 52.7 minutes for the rigid wing to 113.50 minutes for the flexible wing. We also see an increase in the minimum flight times, but this is due to the limitations on the design space, as seen in Fig. 3 for the flexible wing. This increased flight time makes sense, given that flight time is inversely related to average power, which reduces significantly for flexible wings. We see similar trends for the payload mass, with both maximums and averages increasing from the rigid wing to the flexible wing. The minimums remain constant due to the feasibility constraints.

Table 5.

Comparing Flight Times for Different Wings.

	Minimum (min)	Maximum (min)	Average (min)
Rigid Wing	1.00	52.67	2.68
Flapping Only Wing	1.00	24.53	3.36
Pitching Only Wing	1.00	53.51	4.65
Flexible Wing	1.34	113.50	26.11

Table 6.

Comparing Payload Masses for Different Wings.

	Minimum (g)	Maximum (g)	Average (g)
Rigid Wing	1.00	28.42	4.57
Flapping Only Wing	1.00	37.70	5.80
Pitching Only Wing	1.00	38.07	5.50
Flexible Wing	1.00	61.30	7.52

Figure 5 overlays the performances of different wings on top of one-another, from greatest to least average flight time. From this, we can see that the flight times of the flapping only wing (with an average of 3.36 minutes) are only marginally improved from those of the rigid wing. Similar to insect flight on Earth [31], reducing pitching inertial power (which is ignored for the flapping only wing) has less of an impact on the final flight time of the vehicle than reducing flapping inertial power (which is ignored for the pitching only wing). When both inertial powers can be ignored by utilizing a flexible wing, the flight time substantially improves as discussed above. Interestingly, the maximum flight time of the flapping only wing shows a decrease from the rigid wing. This is due to the coupling between the inertial pitching and aerodynamic power, which often cancel each other out [31]. By ignoring the inertial pitching power (as the flapping only wing does), you lose out on this beneficial effect, and the extremely low power (and thus high flight time) designs of the rigid wing are brought closer to the average for the flapping only wing.

Fig. 5.

Performance Comparison.

Determine Attributes of Mission Objective Functions

A swarm of Marsbees would be a modular and more robust system than a single flying vehicle in aiding Martian rover exploration. To be desired by NASA, the system must contribute to one or more of NASA’s four main goals, life, climate, geology, and human, as described in the Background section. To contribute to these goals, the Marsbee’s primary output would be science return (knowledge). Knowledge could then be a simple objective function to maximize and a high-level attribute in a more complex objective function that captures risks, costs, etc.

Knowledge, or science return, can be a challenging attribute to quantify. Sub-attributes of knowledge, which can be quantified in gigabytes of data gathered per year, have been previously identified [32]. These knowledge attributes can be measured through variables such as data collected and papers published. Through scouting missions, the Marsbee would be collecting data on terrain to aid the rover in its mission to explore Mars, where the rover is collecting data. For scouting missions, the Marsbee could increase knowledge of the terrain as well as increase the amount of knowledge gathered through the rover, highlighting the importance of modeling the entire Martian scientific system when optimizing the Marsbee in the future.

As a modular system the Marsbee can use scouting related sensors or non-scouting sensors to gather knowledge on such Martian variables as atmospheric conditions, magnetic variations, and geographical anomalies that are unreachable by a rover. This knowledge would be measured using data, potential papers to be published, etc. and a value would be placed on the gained data using a value model, such as was demonstrated in previous literature for NASA flagship and probe-class missions [32].

In the study performed above, flight time and payload mass were identified as potentially important behavior variables. It is important to note that these variables would not be maximized directly, but instead would feed into a complex objective function where the optimal point may not lie on the pareto frontier identified in Fig. 4. Broadly speaking, flight time was chosen as it may impact the amount of data gathered during scouting missions. Care must be taken as diminished returns will exist as the rover will only be capable of traveling so far, requiring a balance to be found with other attributes.

Payload mass was identified as potentially important due to its relationship with the set of sensors the Marsbee is capable of being equipped with. Different sensors could be used to learn about the Martian atmospheric structure (e.g. winds and pressure), the atmospheric composition (e.g. chemistry), the geology (e.g. land surface structure), the soil’s biochemistry (e.g. land surface composition), etc. Understanding the Martian atmosphere using wind sensors, pressure sensors, and temperature sensors could be potential missions to advance goal two (climate). Path planning and terrain mapping missions can be used to aid the rover in its exploration and to increase its pace. Determining chemical elements and minerals using sensors could also be missions. Each of these potential missions could advance goal three (geology). All of these potential missions would contribute to enabling goal four of conducting human missions on Mars. All of these sensors produce data that represent knowledge that can be related to each of the goals.

V. Concluding Remarks

This paper has identified the models that form a Mars flier system analyzer, and the analyzer’s couplings, parameters, design variables, and behavior variables. The functionality of the analyzer was demonstrated by calculating randomly generated designs and showing the emergences of pareto frontiers. The analyzer formed in this study, paired with an objective function to be fully formed in a future study, would be used in an MDF framework to determine an optimal Mars flier design.

The analyzer could be improved upon in several different ways. For example, incorporating an NS aerodynamics model would open up the possibility to fully analyze flexible wings, rather than making assumptions of the energy-saving qualities of the wings as the current analyzer does. This would add chordwise and spanwise flexibility as design variables. The wing shape could be generalized from a quarter-ellipse to a more generic shape. This would allow different terrestrial wings to be tested and greatly open up the design space of the wing. The accuracy of the added mass and power terms could be improved using cost curves developed using a database of flapping flight vehicles. The battery model could be altered to include more complex, temperature-based battery chemistry, as the operational environment of Mars has extreme temperature swings that should be accounted for when designing a vehicle. Models and analyzers for flight modes other than hover could be developed (e.g. for forward flight, climbing, and gliding) and used together to predict performance of more complicated mission scenarios. All of these improvements would increase the fidelity of the modeling and lead to more accurate optimal designs.