Technical University in Munich (TUM) in 2006.3 After an initial development phase, several system-level ... The matter handling classes were not optimized for performance, and additional ..... ITT âGreenhouse module for space systemâ, see Ref. 17. .... Here the simulation results again agreed well with the online calculator.
44nd International Conference on Environmental Systems 13-17 July 2014, Tuscon, AR
ICES-2014-212
Development Status of the Dynamic Life Support System Simulation V-HAB Claas Olthoff∗ and Jonas Schnaitmann∗ and Anton Zhukov∗ Technische Universit¨ at M¨ unchen, 85748 Garching, Germany
The dynamic life support system (LSS) simulation software Virtual Habitat (V-HAB) has been under development at the Technische Universit¨ at M¨ unchen since 2006. The R MATLAB -based V-HAB software suite dynamically simulates habitat life support systems and their interaction with a detailed human model, as well as the external environment of the habitat. This paper presents the current development status of V-HAB. One of the main areas of work on V-HAB over the past year has been the inclusion of reliability data on the component level to enable the reliability analysis of LSS architectures. Also, parts of the core infrastructure were refactored to provide additional functionality. Within the biological module several new models were developed including microbial trickle filter for waste processing and models of rodents and arthropods. Also the plant model was enhanced to support greenhouse optimization studies. V-SUIT, the spin-off from V-HAB with the capability of simulating space suit portable life support systems (PLSS), has also matured over the past year. A solver for computing liquid flows was implemented. Also, several models of technologies used in the LSS were implemented to depict those components in more detail. The paper describes each of the areas mentioned above in greater detail, presents selected simulation results and concludes with an outlook onto the path ahead for V-HAB.
Nomenclature BP V BV AD C.R.O.P. CDRA DLR CEM ESM EV A F BA F DM GU I HLLC HX ISS JAXA LiOH LSS M CL M T BF OOP P/C ∗ Ph.D.
Back Pressure Valve Baseline Values and Assumptions Document Combined Regenerative Food Production Carbon Dioxide Removal Assembly German Aerospace Center Closed Environment Module Equivalent System Mass, [kg] Extra Vehicular Activities Fixed Bed Adsorber Finite Difference Method Graphic User Interface Harten, Lax, van Leer and Einfeldt (Solver) Heat Exchanger International Space Station Japan Aerospace Exploration Agency Lithium Hydroxide Life Support System Model Confidence Level Mean Time Between Failure Object Oriented Programming Physio-Chemical
Student, Institute of Astronautics, Boltzmannstrasse 15, Building 6 / 2nd Floor.
P LSS S/C TUM V − HAB V − SU IT
Personal Life Support System Spacecraft Technische Universit¨ at M¨ unchen Virtual Habitat Virtual Suit
I.
Introduction
ithin human space flight, the life support system (LSS) is a critical subsystem. Not only is mission W success highly influenced by the reliability, effectivity and stability of the LSS, it is also a major driver for both initial launch as well as resupply mass. 1
For the near future, different scenarios for manned missions are proposed. These include longer-term missions to the moon, missions to near earth objects and one-way or fly-by missions to mars. In order to enable missions further away from earth or long-duration missions, current LSS architectures and technologies need to be further advanced to create more closed-loop, regenerative systems. At the same time, the reliability and effectivity of the systems needs to be increased. For example, a mission to Mars would require an extremely robust LSS whereas a permanent lunar base would require a LSS design with massively reduced resupply needs. As LSS are complex and become even more complex with increased loop closure and regenerativity, the design process is accordingly challenging. Especially in early phases of this process, numerical simulation of LSS can therefore be helpful.2 In order to develop a simulation aiding the design process, a project called the Virtual Habitat (VHAB) was established within the Human Exploration Group of the Institute of Astronautics (LRT) at the Technical University in Munich (TUM) in 2006.3 After an initial development phase, several system-level simulations were conducted throughout the last two years.4 This paper provides an overview about the ongoing development of the V-HAB software and the additional application possibilities of the software enabled through these developments. A.
The Virtual Habitat
The Virtual Habitat Project was initialized in order to develop an engineering tool allowing dynamic simulations of different LSS architectures, enabling optimization of the LSS design in early stages of development. Besides the required dynamic, highfidelity models of each subsystem, it was recognized that the software structure itself is important. To provide a flexible, robust, extendable and readable code, the object oriented programming (OOP) approach was chosen. While a model can posses an internal structure not following the OOP guidelines, each model needs to implement classes providing certain interfaces that can be used to connect the model to the overall simulation. To aid the organization of the different models, they were grouped into different functional modules:
Figure 1. Structure of the V-HAB, reprint from Ref. 3.
• P/C Systems Module: the physical/chemical LSS subsystems are modeled using a component based approach, building the subsystem out of its basic components. The components are modeled based on their chemical, physical and thermodynamical characteristics. • Biological Module: contains plant, algae and other biological models. • Crew Module5 includes a dynamic scheduler and an environmentally sensitive dynamic model of the human physiology Also, a graphical user interface (GUI) was developed to enable linking different subsystems together, as well as different post processing tools that allow examination of the simulated data. 2 International Conference on Environmental Systems
In order to dynamically optimize a LSS design, it was concluded that an approach purely based on the equivalent system mass (ESM) is not sufficient. Therefore, increasing the LSS stability as well as reducing the required crew time for housekeeping and maintenance were also defined as optimization parameters that need to be taken into account.6 Recently, some models covering the ISS air and the water loop were validated against real flight data.7 Also, a thermal solver was implemented and validated against ISS flight data.8 This studies showed that, given a sufficient amount of information regarding the involved subsystems, V-HAB is capable of reproducing the overall system behavior sufficiently. However, due to some software constraints, V-HAB is best suited to simulate “ISS-type” missions in terms of volume, mission duration etc. In order to judge the validity of an overall system-level simulation, a metric called the “model confidence level” (MCL) is used in V-HAB.9 For each model of a system, a value between zero and nine can be assigned (0: only inputs and outputs of the system are defined, 1-3: static model of the system, 4-6: dynamic model, 7-9: (fully) validated model). Each system can be a subsystem to another system. The overall MCL of a system containing several subsystems always matches the lowest MCL of any of the subsystems. Most systems currently used in system-level simulations are modeled dynamically and are, in parts, validated, leading to a MCL of 6-7 for some simulation scenarios. However, as new models are implemented that still have lower MCLs, the overall V-HAB MCL always depends on the actual simulation scenario and is lower for some ongoing studies. B.
Current Developments
As described, it was shown that V-HAB is well suited to conduct system-level simulations for a broad range of missions. However, certain areas were identified where V-HAB would also represent a very helpful tool, but contains certain limitations that would reduce its usefulness. These areas include small-scale LSS with very short time constants and possibly very simple, non-regenerative LSS designs, and also very long-term simulation scenarios with large volumes, e.g. a permanent lunar base. Those limitations are on the one hand related to the implemented models in V-HAB: either models of certain subsystems are sill missing in V-HAB, or their modeling depth is too high or too low. On the other hand, some restrictions exist regarding the V-HAB core. Recently, some of the code in V-HAB was refactored, mainly to overcome two of these issues: 1. The matter handling classes were not optimized for performance, and additional needs regarding matter handling required the implementation of new features. 2. A flexible handling of time steps was required, therefore a new, global timer object was introduced in V-HAB. Naturally, through this several incompatibilities were introduced into the software, requiring additional work on older models. Also, such a refactoring is still required e.g. for the handling of thermal flows and power as well as command/information flows. These functionalities are constantly being implemented as they are needed. Most of the models can be used in with the refactored code, although they do not all yet use components of that framework internally. The different parts in which work on V-HAB is ongoing are separated into three groups each covering specific areas: 1. Infrastructure: ongoing work on the basic software structure of V-HAB. 2. Models: implementation and extension of models in V-HAB representing different LSS sub-systems. 3. System-Level Simulations: the actual application of V-HAB to conduct system-level simulations. The following sections describe some of the changes and additions within each of those areas that were completed in the last two years or are currently being worked on.
II.
V-HAB Infrastructure
The overall goal of the V-HAB core infrastructure is to provide a common, uniform interface for the included models to handle all aspects of the overall simulation. This includes for example a global timer and 3 International Conference on Environmental Systems
scheduler, and also a way to deal with matter. An object of the matter class can store information about the matter composition and properties like temperature, and provides functionalities to calculate different derived values like the specific heat capacity. Other helper classes exist that can, for example, merge two matter objects together and calculate the resulting properties. Besides that, libraries exist that support the developer of a model to, for example, solve matter and thermal flows based on pressure or temperature differences. During the last year, parts of the core infrastructure were refactored in order to simplify the implementation of new models. While every model used in V-HAB can internally use their own logic to calculate the model results, the goal is that each model is eventually connected to other models in V-HAB exclusively through the common interfaces for matter, heat and power as well as spatial/geometric relations and information/sensor data. The same is true for the global timer: instead of managing shorter time steps internally and only synchronizing globally at each global time step, all models will use the new global timer that is capable of handling flexible time steps, separately for each sub-system. A.
Matter and Matter Flows
The matter class, or overall matter handling, was an area of heavy restructuring. This was done to provide higher performance, further extendibility and to implement additional functionalities handling the flow of matter between different matter objects. The matter table class was extended to simplify the addition of new known species and molecules, and to allow different subsystems to reference a separate table, if a specific set of matter properties has to be available that is not relevant for the overall simulation. A performance increase was achieved by implementing some functionalities in a way that the Matlab Code Optimizer can take full effect. For example, accessing the matter properties of each species present in a phase, determining changes in temperature through inflowing matter, and calculating values as the overall molecular mass of a phase were optimized to make use of the Matlab vector and matrix calculation optimizations. To fully handle all aspects of matter flowing between different stores and phases, classes were implemented that represent these matter flows. Each flow path between two phases is called a branch which contains several matter processors that produce pressure drops and can change the temperature. As different phases can have different states (gas, liquid, ...) and can be grouped together in a store with a shared volume, e.g. a liquid water phase with an inflow of matter can decrease the available volume for a gas phase in this store, increasing its pressure. Derived classes can be developed to also take effects of microgravity into account, but are not yet implemented. While phases store matter as absolute masses for each species and derive an overall mass from that, matter flows store an overall flow rate in kilograms per second and retrieve the matter composition from the inflowing phase. A more detailed description of a simulation using this setup can be found in Ref. 10. B.
Timer
The global timer is now capable of fully variable time steps. Each included sub-system can freely set a time step for its regular execution or register specific, one time executions of a function for a certain time in the future. As described in the previous subsection, matter flows are stored in generic objects. For each matter flow, a specificly referenced solver object determines the current flow rate in kilogram per second. While the time step for recalculating the flow rate of this branch is individually set by this solver, the time step in which the actual amounts of matter are moved from the one end of the branch to the other end depends on the size of the stores on these ends. Smaller stores lead to larger differences in pressure for the same flow rate, which in turn forces a smaller time steps for the mass exchange. C.
Thermal
Similar to the matter storage and flow handling, functionality for handling thermal flows is being implemented. For this, classes representing a thermal capacity and a thermal conductor were created. While the matter classes require additional, derived types for the different states of matter, the thermal classes need to be able to handle the different transport mechanisms of thermal energy. For enthalpy flow, the information already available and provided by the matter flows is used. For
4 International Conference on Environmental Systems
radiation and conduction, separate branch types for those thermal flows were defined. Currently, work is being done to allow each element in V-HAB that deals with matter (store, processor etc.) to also represent a thermal node, either a capacity or a conductor with the according properties. Other than for the matter solver, it was decided to create one, global solver that handles all thermal flows as such a solver was already implemented as a library for V-HAB8 and correlated with ISS flight data. On initialization, the solver retrieves all required information about heat capacities, resistances etc. from the different nodes to access those values more quickly. On execution, all parameters that can change throughout the simulation are updated and the V-HAB thermal solver is called with this information. At the time, first validation trials with this approach are being conducted. D.
Matter Flow Solver
Two simple ways to set flow rates in V-HAB exist, a “manual” control object for setting a fixed flow rate for a branch, and a solver simply equalizing pressures or masses between phases. Additionally, different solvers exist that try to calculate realistic flow rates. The currently available solvers only support gaseous phases and implement an iterative approach and one based on the hydraulic diameter. However, they do not suffice in terms of stability, validation and uncertainties about the calculation error. Therefore, in order to properly assess the flow rate between two phases, a new, physics based solver was implemented using the finite difference method (FDM solver). Its design is based on solving the Euler equations for inviscid, friction less gas flow with a finite difference scheme with adaptive time steps. At the start of each simulation, the desired fineness of the discretization and maximum error for each branch must be set. The program then automatically constructs two discretizations of the branch, one for computing the flow rate for each time step and a finer one which is used for error estimation. Time steps are calculated automatically to ensure the stability of the simulation. On the one hand it is possible to use the solver on a coarse discretization if a relatively fast computation is necessary. On the other hand it can be used for fine discretizations if a low error tolerance is important, albeit at the cost of longer runtimes. With the FDM solver, basic modularity and functionality of the V-HAB LSS simulation remain unchanged while giving the user the ability to adapt the accuracy and speed of the simulation to his or her specific needs. Figure 2 shows the tank masses and flow rate of a system of two tanks connected by a pipe with a length of 1m and radius of 1 cm are displayed. The initial pressure difference is 2 bar. The FDM solver was used with a 30 cell grid.
Figure 2. Plots of tank masses (left) and the flow rate between those tanks (right) calculated with the FDM solver.
Besides the described FDM solver being able to calculate flow rates for gases, another solver for compressible liquids is in development. Using an approximate Harten-Lax-van Leer-Contact (HLLC) Riemann solver for the solution of the fluxes between numerical cells, this solver is capable of calculating shock wave effects in liquids at high pressure differences as well as continuous liquid flow effects. The calculation of flow rates and pressures is implemented for systems utilizing components with pressure influence and also for tanks
5 International Conference on Environmental Systems
containing both liquid and gas phases. Components influencing temperature so far can not be calculated correctly, only the influence of temperature changes from the transported matter itself is implemented. As this part of the program is still being developed, the final functionality is not yet determined. This also applies to the influence of diameter changes of e.g. pipes which is implemented but has not been validated yet. Ongoing work therefore focuses on the implementation of diameter and external temperature changes as well as validation of the solver itself.
III.
Modules and Models
During the last year, many new components were introduced into V-HAB. As described in section A, the P/C Module of V-HAB uses a component-based modeling approach. Technologies are created from common, basic components that can be extended to suit some more specific needs. New components were included in the simulation that allow the creation additional technologies, and existing components were upgraded, leading to more sophisticated versions of existing technologies. Within the biological module, several models were newly introduced and the plant model was upgraded. The following subsections describe some of these new functionalities within the Biological and P/C Modules. A.
Biological Module
From the beginning, plants and algae were a part of V-HAB. In the last years, technologies and models were included to allow the simulation of additional biological elements. On the one hand, new or more detailed models of non-regenerative, but simpler P/C systems were implemented. On the other hand, models of different animals were and are being developed and integrated. This enables V-HAB to also simulate small scale LSS for plants and animals. 1.
Arthropods Model
Several concepts to use insects to supplement the food chain exist, both for terrestrial as well as space applications. On the one hand, arthropods could be used as food for animals later consumed by humans, on the other hand many insects can be consumed directly if properly prepared. To allow the examination of such scenarios in V-HAB, a model depicting larger populations of arthropods is being implemented. Focus of the model is reproducing the environmental factors influencing biomass production of such populations. These factors range from light, temperature, humidity, day/night cycle and atmosphere composition to additional, manually triggered stress influences and food including shape and nutritional value. Different parameter sets can be used with the model to represent different species. Effects of trace gases and environmental pollutants will not yet be implemented. The model does represent the whole population by different groups, each representing a certain amount of individuals of a specific development status, determined by weight. Depending on the desired time step and the overall life span of the simulated species throughout each development status, the amount of groups is determined on initialization and therefore the simulation resolution with respect to time. For example, for the black soldier fly, the total life span is 1018 hours (from the status egg to adulthood and death). Each time step, a certain amount of individuals is moved from each group to the next, and depending on the development status represented by each group, a certain amount of individuals dies. The groups representing adult, fertile individuals can, in turn, be sources for new individuals in the first group (eggs). All described rates are influenced by the previously described environmental conditions whose impact or requirements can depend on the development status of the animals in that group, in case of the nutritional needs for example. As each group represents individuals of a certain weight, empirical data is required to configure the model accordingly. Where such data is not available yet, the weight is interpolated between known values. This behavior is incorrect as generally, arthropods grow mainly after moulting. As the amount of individuals and their individual weight is known for each group, the oxygen consumption and carbon dioxide production can be calculated based on the total weight. At each time step, individuals can be removed from any group. Depending on the required type of biomass (adult, larvae), the needed amount can be removed from the according group in a regular interval. Figure 3 shows some simulation results from this model. When a certain population size is reached, a constant amount of adult larvae is removed regularly to keep the overall population size constant. For each 6 International Conference on Environmental Systems
Figure 3. Plot of simulation results with the arthropods model in V-HAB.
Figure 4. Comparison of the simulated concentration of different substances in C.R.O.P. with the measured data.
7 International Conference on Environmental Systems
larvae and day, 100mg grams of chicken is provided as food. Additional use of some species of arthropods could be made in monitoring. For example, daphnia are very sensitive to environmental pollutants. This dependency could be implemented in V-HAB to represent biological sensors capable of providing additional information, which is planned for the future. 2.
C.R.O.P.
The Combined Regenerative Food Production (C.R.O.P.) concept is currently being investigated at the Institute of Aerospace Medicine of the German Aerospace Center (DLR) in Cologne, Germany.11 It is a biological trickle down water filter that is able to microbiologically recycle organic nutrients from waste water and biological waste as inedible biomass to efficiently re-utilize those waste streams. It produces a nitrate-rich fertilizer solution that can be provided directly to the plants in the LSS. Numerous experiments have been and are being conducted by DLR in order to fully understand and characterize the process. The filter consists of a tube containing porous volcanic rock (red lava) providing a large surface for aerobic bacteria as well as anoxic environments for anaerobic bacteria due to their porosity. The filter is incubated using common garden soil and a constant water stream trickles down through the tube to transport the waste water through the filter. The aerobic nitrifying bacteria within the filter can subsequently start to break down e.g. the urea in the waste water. An advantage of this system is that it can be stopped and restarted when needed.
Figure 5. Intake and outputs of trickling filter in batch modus.
In a first step, a model was implemented covering the nitrate cycle.12 The filter contains many different species that are not yet fully identified, therefore focus was put on the more important processes. The model is based on enzyme-catalyzed reaction kinetics for the fundamental microbiological reaction chain from urea
8 International Conference on Environmental Systems
to nitrate. Each reaction (with the exception of the auto-ionization of ammonia, which is not enzymecatalyzed) is modeled with a general inhibition reaction scheme, allowing different forms of inhibition. The model is validated trough model parameter fitting (reaction rate constants) by comparing the time series solution of the model to the experimental data. Figure 4 shows a comparison of simulated and measured data. Finally, a basic assessment of the models behavior under different operating conditions was conducted. A model of a reverse osmosis filter, which can filter ionic substances (ammonium, nitrite, nitrate) from non-ionic ones (urea, ammonia) and separate them into a highly concentrated solution, was used to simulate sequential batch processing and quasi steady-state operations of the filter. Several sets of exemplary operating frameworks were analyzed. It was found that with a daily turnover (putting solution through the reverse osmosis filter) of the entire tank (30 liter) a throughput of around 30 g of urea per day (about the average daily human output) is feasible, see figure 5. However, the results have a predictive character and need to be verified through the experiment. In general, the technology is considered as very promising and has many advantages. DLR, Cologne, continues the development and currently extends the experiments by processing of bio waste (cellulose, plant waste) as well as running the system in batch mode. The obtained data will be used to improve and correlate the model further. Also, it is planned to extend the model to cover the cellulose cycle. 3.
Plant Model
V-HAB contains a plant model based on the modified energy cascade model as described in Ref. 13 (BVAD). It can be adapted by a parameter set to simulate the metabolism of 21 different plant species. Recently, the code of the plant model in V-HAB was refactored to fully incorporate the object oriented programming approach. Due to its new structure, it can be used in a more flexible way opening up additional possibilities to, for example, optimize the plant setup to shorten the start-up phase, also e.g. after failures. The modular structure allows setting up different plant chambers with different properties (size, lightning, soil or hydroponics) containing different plants/generations that can either be placed within a dedicated, separate module or within the overall habitat sharing the same atmosphere as e.g. the crew and the P/C subsystems. Furthermore, new functionalities were implemented in the model. For some plant species, hydroponic planting can now be used. Also, plant density can now be taken into account. Ongoing work includes correlating the plant model with additional, empirical data both from literature and from experiments, for example data obtained from the NASA Biomass Production Chamber.14 The results from the simulation match the NASA experiments with the hydroponic factor for tomato, wheat, potato, lettuce and soybean.
Figure 6. Left side: total edible biomass productio. Right side: total CO2 consumption. Red lines: one generation of each plant species. Green lines: two generations of each species.
Plant density has been added as a new factor to the simulation based on the Bleasdale’s simplified equation.15 Due to lack of reliable data, the plant density parameter has been correlated only for tomato using data from experiments with direct seeded tomatoes,16 which were confirmed with the Bleasdale’s equation. The approach has been implemented in the plant model and can be used in future for all species in case the reliable data can be found. Tomato serves as a prototype for the density sensitivity. One case study done with the new model focuses on optimizing the steady-state growth rate of (edible) 9 International Conference on Environmental Systems
Figure 7. Total edible biomass production of a reenhouse setup optimized on reduced start-up and constant steady-state production.
biomass to provide a constant carbon dioxide consumption and food production and is based on the ESA ITT “Greenhouse module for space system”, see Ref. 17. Figure 6 shows the results of a setup with two plant chambers, one containing wheat, soybean, potato and rice cultures and the other containing lettuce and beet. The red lines show the results when one generation of each plant species is grown. To reduce the oscillations of the production rates, two generations of each species were planted on the same area with an according time delay. Using this approach, the green house was optimized leading to an biomass production rate as shown in figure 7. In comparison with the scenario based on BVAD data, a 30% reduction in the required planting area was achieved. Present development of the plant model is focused on the implementation and correlation of sensitivity towards the spectral properties of the light. The work is performed in close cooperation with NASA KSC. A series of experiments is conducted to characterize the effect of light with different spectra on plant growth and performance. Obtained data will be used for model correlation. It is expected that such model sensitivity will improve the greenhouse optimization studies allowing variation of light source types. 4.
Rodents Model
As described in more detail in a later section (IV.C) a simulation of a small, non-regenerative LSS (AEME) containing mice was created in V-HAB. In this simulation, the mice were represented by a black box using predefined rates for CO2 and H2 O production and O2 consumption, one set for the sleeping period and another one for active periods. To depict the rodents contained in the LSS more precisely, a mathematical model based on empirical data was developed which can represent larger populations of mice.18 An extensive literature research was conducted, collecting data for different strains and both genders. In order to define a time-dependent growth rate for the mice, it was not only necessary to differentiate the data by gender, but also by a weight category (thin, normal, thick, obese) assigned to each strain. Also, the growth rate depends on the age of the mice, which are growing exponentially in the first few weeks followed by a decreasing growth rate, and reach their final weight after around 16 weeks. Based on the weight of the mice, the oxygen consumption is calculated, additionally taking other parameters as temperature and the day/night cycle (activity) into account. Carbon dioxide and heat production are derived using the respiratory quotient, the O2 consumption and, in case of heat, the caloric equivalent. Additional values as food intake and the water cycle are implemented as well, however, less reliable data was available. Figure 8 shows some of the results produced with this model. Ongoing work aims on creating the according numerical model to include this module in the overall AEM-E LSS simulation.
10 International Conference on Environmental Systems
Figure 8. Left plot: Weight development of mice for the different weight categories. Right plot: oxygen consumption for different temperatures and the day/night cycle.
B.
P/C Module
As mentioned in the previous sections, the time step in V-HAB was made variable to, for example, allow the simulation of smaller systems and volumes with shorter time constants. This is especially important for the Virtual Suit Project (V-SUIT)19 which uses V-HAB to simulate personal life support systems (PLSS). In these simulations, the available air volume is very small compared e.g. to the ISS. Therefore, smaller time steps are required. Models of different components used within the PLSS and other small-scale LSS were upgraded or implemented to allow a more detailed depiction of these sub-systems. While these more sophisticated models lead to a longer runtime of the simulation, they are necessary in order to ensure that the simulated technologies meet the higher requirements of those more sensitive LSS environments. 1.
Fixed Bed Adsorber
To enable a more detailed modeling of CO2 removal systems, a more in-depth model of a fixed bed adsorber using zeolite was implemented, see Ref. 20. This is important for both the ISS system-level modeling efforts as well as the ongoing work of modeling further technologies in detail to conduct comparison studies. In the past, flow rate simulations of a CO2 removal system being developed at JAXA were used to to validate the flow rate solver of V-HAB.3 Data from the same system was used in the development of the adsorber model in V-HAB. The model implements both thermodynamic as well as kinetic aspects of the adsorption process and can therefore predict the behavior of the filter with respect to the properties of the inflowing gas and the bed itself as well as the geometric properties of the filter. Two different solving algorithms are provided to allow either a more precise or a faster simulation. A higher-dimensional thermal model for the adsorber bed itself is not yet implemented. Also, additional matter properties need to be researched and implemented in the future to allow the simulation of adsorption materials other than zeolite and to enhance the depiction of multi-component adsorption, which is to date only included for CO2 and H2 O. Figure 9 shows a comparison of measured and simulated CO2 breakthrough curves. It can be seen that the simulation fits the experiments well, as the properties of the filter are known precisely. Within the validity ranges of the model, other filter dimensions should produce acceptable results as well, as the model was compared to experimental data with varying flow rates. Also, different types of zeolite can be used as the different isothermes are known. However, parameters not known as e.g. the geometry of the zeolite chunks can have an impact. Also, different filter geometries might require additional work to depict the according gas flow paths through the adsorption material. Additionally, the model is for example limited in temperature and pressure ranges due to the usage of the ideal gas equation, and the flow rate is limited as high gas velocities are not supported.
11 International Conference on Environmental Systems
Figure 9. Comparison of measured and simulated CO2 breakthrough curves for two different cases (varying in flow rate). Red: simulation, blue: measurement. While the difference between the model and the experimental data in the case shown on the left plot (higher flow rate) is noticeable, the simulation results are still acceptable as V-HAB is intended to be a tool for rough estimations in earlier phases of the LSS design. It is assumed that the deviation is mainly due to inaccurate geometric properties set for the model. The differences in the results on the right side are assumed to be negligible.
2.
Amine-Based Vacuum Swing Adsorption Unit
In the V-SUIT simulation, NASAs portable LSS 1.0/2.0 model is being implemented. Therefore, a model of the Rapid Cycle Amine (RCA) subsystem, which removes CO2 and H2 O, was required.21
Figure 10. Original test data of the RCA.
The developed model contains the two adsorption beds and switches between beds as soon as an output threshold of 900 Pa CO2 partial pressure is reached. The filter model splits the adsorption bed into several nodes and uses an implicit solving algorithm to calculate the gas kinetics. The adsorption rate of carbon dioxide is determined using the Toth isotherm model, while the water vapor adsorption rate is based on the Freundlich adsorption isotherm. The pressure drop through the filters is calculated using the formula of Blake-Kozeny. Figure 10 shows test data from the actual PLSS 1.0 system. Figure 11 shows simulation results generated
12 International Conference on Environmental Systems
Figure 11. Simulated data of the RCA model.
with the new RCA model. It can be seen that the shapes and trends of the lines representing CO2 (Fig. 10, red and green lines; Fig. 11, left plot) and H2 O (Fig. 10, blue and yellow lines; Fig. 11, right plot) inputs and outputs match the original data from the PLSS 1.0 system fairly well. There are however subtle differences in the shapes of the individual traces, even when ignoring the measurement induced jitter of the test data. These differences can be attributed to the fact, that the test data are from a full PLSS system prototype with actual sensors positioned at the suit volume simulator inlet, whereas the simulation only included the RCA with constant input mass flows and virtual sensors positioned directly at the RCA outlet. Further investigation into these effects is ongoing. 3.
Heat Exchanger
As temperature and humidity control are integral parts of every LSS and especially the PLSS mentioned in the previous section, a more in-depth, parameterized model of a heat exchanger (HX) was developed, see Ref. 22. Figures 12 and 13 show simulated results using the new HX.
Figure 12. Example results from the new HX model, temperatures of the fluids through a parallel and a counter flow heat exchanger.
In a preliminary step, the most important and widely used types of heat exchanges were identified. A model was then created supporting not only parallel-, counter- and crossflow heat exchangers, but also certain types of shell and tube heat exchangers with different amounts of inner and outer passes. To calculate the overall heat exchange coefficient of the exchanger, the Effectiveness method was used, also taking into account the geometry and the material of the heat exchanger. Therefore, only knowledge of the geometric type of the heat exchanger and the materials used need to be known to include the model in ones own systems and technologies. Additionally, again depending on the type of HX, the pressure drop through the component is calculated
13 International Conference on Environmental Systems
Figure 13. Crossflow heat exchanger setup. The planes respectively represent fluid 1 (upper plane) and 2 (lower). Fluid 1 flows along the Xi axis whereas fluid 2 flows along the Eta axis. In the plot on the right, the fluid 1 represented by the upper plane is perfectly crossmixed, i.e. only changes its temperature along its flowing axis, but is mixed and has an equal temperature along the flow axis of fluid 2.
using first order fluid dynamics calculations. This was deemed acceptable due to the relatively simple geometry of the heat exchangers. Using the finished, generic HX simulation, a model of the heat exchanger used in the PLSS 1.0 testing was created and the results compared to published, steady state test data. Without any modifications to the underlying calculations, the model predicted the test results for the outlet gas temperature within 1.4 K and the outlet water temperature within 0.09 K. The model was also compared to the results of a calculator for heat exchangers found on the website (http://exergyllc.com/calculator.php) of the vendor that produces the HX used in PLSS 1.0 testing. Here the values were nearly identical. Since the solver for liquids was not completed at the time of writing, the pressure drop was only calculated for the gas side. Here the simulation results again agreed well with the online calculator. Due to limited time, the dynamic behavior of the model was only demonstrated by simulating step changes in inlet temperatures. The comparison of the results to actual test data is future work. As described in section II.D, a matter flow solver for liquids is currently being developed. Together with properly modeled valves, pipes and pumpes, this model will enable V-HAB to simulate liquid cooling garment loops within LSS and especially the NASA PLSS 1.0/2.0 including a dynamically calculated flow rate.
Figure 14. Increase in heat rejection during inlet temperature increase and constant BPV position.
14 International Conference on Environmental Systems
Figure 15. Inlet and outlet temperature traces during inlet temperature increase and constant BPV position.
4.
Space Suit Water Membrane Evaporator (SWME)
Another project associated with V-SUIT was the development of a simulation of the space suit water membrane evaporator (SWME).23, 24 This component serves to cool the water loop which acquires thermal energy from the suited crewmember via a liquid cooling garment and the space suit avionics via a cold plate. The water loop also cools the ventilation gas stream before entering the suit volume through a heat exchanger (modeling described in the previous section). The SWME is based on porous hollow fibers that allow the evaporation of water to vacuum, thereby cooling the water flowing through the fibers.25 This evaporative heat transfer as well as the convective heat transfer from the flowing water to the fiber walls was modeled for all flow regimes through the fibers (laminar, transition, turbulent). The temperature and pressure dependence of the associated properties of water were taken into account. The hollow fiber bundle in the SWME is in a sealed enclosure with a back pressure valve (BPV) controlling the pressure inside the enclosure. By regulating the pressure around the Figure 16. Decreased heat rejection due to ris- fibers the rate of evaporation can be changed to achieve differing environmental pressure. ent water outlet temperatures. The enclosure, BPV and the surrounding vacuum was also modeled. The fiber bundles are modeled within V-HAB as one element (flow processor) that performs the necessary heat transfer calculations. This may be expanded to a multi-element model in the future to improve model fidelity. To verify the SWME performance, its sensitivity to varying inlet temperatures as well as different fiber tortuosities, inlet mass fluxes and external pressures were examined.26 The simulations produced the expected results: increased heat rejection at increased water inlet temperatures (Figures 14 and 15), decreased heat rejection for greater external pressures, due to increased water vapor pressure inside the SWME (Figure 16), and decreased heat rejection for a larger membrane tortuositiy Figure 17. Decreasing heat rejection due to higher membrane turtuosity.
15 International Conference on Environmental Systems
factor (Figure 17). The SWME simulation results were compared to published test data.27 The simulated heat rejection for an open valve simulation is slightly higher than the results of the real life experiments. (Figure 18, left plot) The average difference in heat rejection, however, constitutes less than 5% and might be reduced by adjusting simulation parameters, like the membrane’s tortuosity factor. The simulation results for the heat rejection at back vapor pressure also represent the results of the real life SMWE experiments well and only slightly differ for inlet temperatures greater than 30 Degrees Celsius. (Figure 18, right plot) The average difference in magnitude for those inlet temperatures is less than 6%.
Figure 18. Left plot: Comparison of open valve simulation data with experimental test results. Right plot: Comparison of heat rejection at back vapor pressure data from simulation and experimental test results.
Despite the good results of the simulation, the SWME simulation system still has some limitations. To improve the simulation, the following approaches might be considered: consider additional physical processes like the heat conductivity through the membrane, the heat convection from the membrane to the water vapor and the pressure drop inside the vapor phase due to the matter flow inside the phase.
IV.
System-Level Simulations
The V-HAB simulation software was developed mainly with the goal of conducting system-level simulations. As these top-level effects are produced by the interaction of the involved subsystems, it is important that the models of those subsystems are sufficiently detailed to reproduce their relevant behavior. If a sufficient amount of experimental data is available, technologies can be represented by top-level, black box models reproducing their behavior. However, for different technologies included in V-HAB, such data was not available. For those, a component based modeling approach was chosen (see secion I.A). This means that in a typical V-HAB setup, effects can propagate up from the components through the technology and subsystem level to the overall system level. This enables the user to examine various scenarios, for example conducting failure analyses simulating the effect of a broken component. Another area of interest is the question how a technology or subsystem introduced in an existing system would change the overall system behavior, stability and reliability. Ref. 28 describes the simulation of a liquid cooling garment in detail, conducted with V-HAB. A.
ISS System Level Simulations
Last year, results of system level simulations of the ISS done with V-HAB were compared to actual ISS flight data.7 Work on this project is ongoing. On the one hand, the models used in this simulation were updated so they can be used in the new V-HAB environment and some additional funcitonality was added.29 On the other hand, further studies are conducted, in parts at NASA JSC, that use these validated models to simualte a Deep Space Hab. As an example, figure 19 shows the semantic of the 4BMS CO2 Removal System model implemented in
16 International Conference on Environmental Systems
Figure 19. Left side shows the semantics of the CDRA model in V-HAB. The right side shows the differences in CO2 flow rate between the old an the new version of the model. Due to a difference in the CO2 flow rate calculation, the spikes in the plot produced by the new version are much higher (not shown in the figure) but also much more narrow, which in total leads to a similar mass of transported CO2 .
V-HAB, resembling the Carbon Dioxide Removal Assembly (CDRA) on ISS, and the CO2 flow rates out of the beds. It can be seen that there is a certain discrepancy between the original simulation data (red line) and the newly implemented system. A similar disagreement can be seen with the bed temperatures during the desorption cycles as shown in figure 20 on the left side. The right side in this figure shows the partial pressures of CO2 in the atmosphere for both versions of the model. Further tests have to conclude if these differences do make sense and were introduced due to the inclusion of some additional functionalities and dependencies, or if the new version of the model still has some implementation errors.
Figure 20. During desorption, the bed temperatures disagree. From literature, evidence was found that the temperature from the newer model is actually correct. The right side shows the eventual difference in CO2 partial pressure produced by the old and the new version of the model.
In a next step, it is planned to make this simulation fully dynamic, i.e. relying on dynamic, pressure dependent flow rate calculations in all technologies. This would allow to partly optimize the different subsystems to better match flow rates and buffer sizes. B.
JAXA CO2 Removal System
As described earlier, a fixed bed adsorber (FBA) model was created simulating a zeolite CO2 filter. Currently, work is ongoing that implements a more detailed model of the overall CO2 removal system and integrates the new FBA model. Validated pressure drop simulations with the according dynamic calculation of flow rates were already conducted for a subscale version of this system. Therefore, the model combining both the components producing correct pressure drops, and the new FBA component, should be able to reproduce the system well. With that model, simulation and comparison studies can be run to analyze the performance of the model further. Figure 21 shows an example of these pressure drop calculations. In a subsequent step, simulations can be created to introduce this detailed model of the JAXA CO2 scrubber in the previously described ISS system level simulations and examine the changes and effects this system would have on the overall ISS atmosphere. Also, work is ongoing to model other two systems typically involved in air revitalization, electrolysis and a sabatier reactor, in more detail. This would allow comparison studies e.g. between the current ISS air revitalization loop and newly developed systems.
17 International Conference on Environmental Systems
Figure 21. Comparison of the pressure drops for the same flow rate through two simulated systems (not resembling the actual system). The system on the right was designed to produce less pressure drop, e.g. by using different valves. Pressure drops were either experimentally measured or data from the component’s producers was used.
Figure 22. Semantics of the AEM-E simulation model.
Figure 23. Left plot: total masses in Cabin. Right plot: pressure in O2 tanks.
18 International Conference on Environmental Systems
C.
AEM-E
As mentioned, some of the changes were introduced to be able to simulate small-scale LSS, partly relying on very simple mechanisms for controlling e.g. the atmosphere. One example is the “Rodents on Dragon” project where simulation studies were conducted with V-HAB. Figure 22 shows the semantic structure of the simulation model. The oxygen tanks on the right are redundant and four in number. The valves for each oxygen tank are opened periodically if the oxygen levels become too low, and an orifice plate after each valve regulates the flow rate. The mice within the capsule produce CO2 and H2 O and consume O2 depending on a day and night cycle. The flow rate through the LiOH filter is held on a constant value, and the model calculates an extraction rate for H2 O besides CO2 , based on some simplified assumptions. The blower pushing air through the SiO2 beds is controlled through a sensor, switching on when the humidity reaches 60% and switching off at 40%. Figures 23 and 24 show results of the simulation. The oxygen pressures oscillate according to the opening and closing of the O2 supply tank valves. As the pressures in those tanks drop, the duration the valves need to be open increases. Because the O2 pressure in the tank is lower, the flow rate through the orifice plate decreases and the upper O2 threshold for closing the valve again is reached later. It can be seen that the overall pressures oscillate mostly in accordance with the oxygen partial pressure. After day seven, the oxygen supply drops below a critical value and the O2 partial pressure, and with it the overall pressure, starts to drop.
Figure 24. Left plot: CO2 pressure in Cabin. Right plot: total pressure in the different volumes.
As the flow rate through the CO2 removal system is held constant, the carbon dioxide level rises and drops according to the day/night cycles of the mice. Also, as the filter becomes fuller, the CO2 partial pressure increases slightly in average towards the end of the simulation. As the filter is sufficiently large, this effect however is small for CO2 . However, this saturation effect within the LiOH filter happens faster for H2 O (no figure shown). Within the first six days, the relative humidity levels in the cabin follow the same pattern as the CO2 levels and oscillates around 38%. After six days however, the relative humidity in the system starts to rise. After another twelve hours, the upper threshold of 60% of relative humidity is reached and the SiO2 system is activated until the humidity drops below 40%. This effect can be seen in the oscillations of the total pressures, starting around 6.5 days. D.
Failure Layer
The development of ultra-reliable LSS is especially relevant for missions further away from earth, as resupplying lost consumables or spare parts becomes increasingly expensive. As V-HAB composes many 19 International Conference on Environmental Systems
technologies from their underlying components, analyses are possible that use the components mean times between failures (MTBF) to trigger such failures in single components to observe the overall effect on the systems stability due to this error. Based on how severe the failure impacts relevant system parameters, the priority of the repair task can be determined.
Figure 25. Example of a fault tree analysis (FTA) of low total cabin pressure.
An extensive study, see Ref. 30, was conducted including a literature research to create a database of components including their typical failure modes and frequencies, and repair times. Different methods for modeling and analyzing reliability were compared, for example the fault tree analysis (FTA) and the event tree analysis (ETA). Subsequently, the collected data was applied to the components of a typical air revitalization system. The resulting tables map components and their properties to the systems using them. In a final step, a FTA and an ETA were presented for some important error cases to demonstrate the principles of those methods. Figure 25 shows an example tree, generated for the Air Revitalization System covering the case of a low total pressure. Following this, it is planned to conduct initial studies with V-HAB, generating data about the severity of certain failures of specific components within technologies. For this, it is necessary to include the relevant data in the component library used in V-HAB. Using the collected and simulated data, the required inputs for a fault analysis for a given scenario can be generated including the properties of the different components used and also the hierarchy between the components, sub-systems/technologies and systems. Therefore, an FTA or ETA can be conducted prior to the actual simulation runs of the given scenario. Also, the according analyses could be run for different phases of the simulation, if e.g. the configuration of the components of a system change. As a result, during a full-scale simulation a stochastic process can trigger component failures based on MTBFs and other values. Using the fault analyses results, the dynamic scheduler can estimate the impact of the failure and accordingly set a priority for the repair task.
V. A.
Conclusion
Discussion
In many areas, V-HAB was extended and advanced throughout the last two years. Some of these changes in more specific areas or projects were described in previous papers. This paper gives an overview of the changes in core areas that were implemented due to different reasons: on the one hand, the scope of V-HAB was extended throughout the last years, for example to also cover short-term, small volume scenarios but also very long-term, large volume missions. On the other hand, the existence of several validated, system-level simulations showed that with V-HAB, dynamic simulations using a component based, bottom-up modeling approach reflecting some of the overall system behavior are, in principle, possible. Therefore, a wider set of 20 International Conference on Environmental Systems
modeled, ready-to-use components and technologies is now desired to be available within V-HAB to allow additional simulation setups. The most basic changes were done in some areas of the core infrastructure of V-HAB, changing the handling of matter in different phases and flows of matter between those phases to provide an even more uniform handling of matter. Also, the global time step handling was refactored to allow flexible time steps. New solvers are being implemented that calculate flow rates for gaseous and liquid matter flows between different phases, including error estimations of the solution that was found. Additionally, functionality covering thermal flows is being implemented. Different new components and technologies were introduced in V-HAB. Those range from biological models representing arthropods or a biological trickle filter to the updated plant model. Within the P/C Module, several new models were added, including a fixed bed adsorber model, an amine based CO2 removal technology and a heat exchanger. These new technologies allow the depiction of additional missions in VHAB in more detail, for example a small-scale animal LSS or NASAs PLSS 1.0/2.0. Using those components and others currently being developed, new system-level simulation studies were conducted or are ongoing. B.
Outlook
As described, some core areas of V-HAB were refactored, which led to several incompatibilities. After those issues are resolved and all models are consolidated into the new structure, several additional tasks are planned: • Extending the biological models, for example extending the biological trickle filter to also cover the breakdown of fibers. • Implementing a system-level simulation of the PLSS 1.0/2.0 including the new detailed models, while still achieving an acceptable simulation runtime. • Implement top-level models of some systems to enable simulations with a much sorter runtime to e.g. simulate long-term missions. Generally, it is planned to implement several versions of important models with different modeling fidelities. The same scenario can then be simulated using the different model versions, allowing the examination of the potential benefits of the high-fidelity model versus the simpler and likely faster simulation with the low-fidelity models. Eventually, this would allow using simpler models for early planning phases with many possible configurations and optimization iterations. In later phases, where some selected configurations have to be analyzed in more detail, the higher-fidelity models can be used. • Eliminate some known issues that lead to increased simulation runtime, instabilities, or non-intuitive object structures. • Adding additional ways to visualize the simulated LSS and plot the calculated results. • Implementing a way to configure and run simulations scenarios without having access to the software itself, preferably though an internet based technology. This would allow providing V-HABs functionality to others while not giving away access to the models themselves, as this is sometimes not possible. • Implementation of additional ways to run V-HAB in batch mode, i.e. to run the same simulation several times with varying input parameters.
Acknowledgments The authors would like to thank the many students that contributed to V-HAB and the models and projects presented here through their bachelors and masters theses. Last but not least the authors thank Professor Ulrich Walter for his continuing support of the V-HAB project.
21 International Conference on Environmental Systems
References 1 Jones,
H., “The cost and equivalent system mass of space crew time,” SAE Technical Paper (2001): 2009-01-2359. H., “Planning Dynamic Simulation of Space Life Support,” SAE Technical Paper (2009): 2009-01-2493. 3 Czupalla, M., “The Virtual Habitat - Integral Modeling and Dynamic Simulation of Life Support Systems,” PhD thesis, Technische Universit¨ at M¨ unchen, Verlag Dr.Hut, ISBN 978-3-8439-0305-9, Munich 2011. 4 Czupalla, M., Zhukov, A., Mecsaci, A., Deiml, M., Beck, M., “Dynamic Life Support System Simulations with the Virtual Habitat,” In Proceeding of the 41th International Conference on Environmental Systems, Portland, USA, ICES-5038, 2011. 5 Schnaitmann, J., Zhukov, A., Hager, P., Klaus, D. and Czupalla, M., “The status of the environmentally sensitive dynamic model of the human physiology used in the V-HAB LSS simulation,” AIAA 2012-3467, 42nd ICES, 2012, San Diego. 6 Czupalla, M., Dirlich, T. and Bartsev, S.I., “An approach to LSS optimization based on equivalent system mass, system stability and mission success,” SAE Paper No. 2007-01-3222, 37th International Conference on Environmental Systems, 2007 7 Ploetner, P., Roth, C., Zhukov, A., Czupalla, M., Anderson, M., Ewert, M., “Status of the Correlation Process of the V-HAB Simulation with Ground Tests and ISS Telemetry Data,” AIAA 2013-3476, 43nd ICES, 2013, Vail. 8 Zhukov, A., Roth, C., Ploetner, P., Czupalla, M., “Simulation of the Temperature and Humidity Control System of International Space Station in the Virtual Habitat,” AIAA 2013-3454, 43nd ICES, 2013, Vail. 9 Czupalla, M., Hager, P., Hein, A., Dirlich, T., Zhukov, A., Pfeiffer, M. and Klaus, D., “Model Confidence Level - A Systematic Metric for Development of a Virtual Space Habitat,” SAE Paper No. 2009-01-0208, 39th International Conference on Environmental Systems, Savannah, USA, 2009. 10 Olthoff, C., “Development Status of V-SUIT - The Virtual Space Suit simulation software,” AIAA 2013-3481, 43nd ICES, 2013, Vail. 11 Hauslage, J., Bornemann, G., Waer, K., Tonat, T., Hemmersbach, R., Bohmeier, M., M¨ oller, R., and Anken., R., “Modification of biological wastewater treatment processes: Using trickling filters to produce fertilizer from urine,” Submitted for publication in Ecological Engineering, 2014 12 Tertilt, G., “Development of a model of a microbiological trickle filter for bioregenerative life support systems,” Diploma Thesis, Technical University Munich, RT-DA-2013/01, 2013 13 Hanford, A., “Advanced Life Support Baseline Values and Assumptions Document,” NASA, Johnson Space Center, Houston, TX, Tech. Rep. CTSD-ADV-484, 2002. 14 Wheeler, R. M., “Nasa’s biomass production chamber: A testbed for bioregenerative life support studies,” Adv. Space Research Vol.18 pages 215-224, 1996 15 Willey, R.W., “Heath S.B.: The quantitative relationships between plant population and crop yield,” Advances in Agronomy Volume 21, pages 281-321, 1969 16 Nichols, M. A., Nonnecke, I. L., and Phatak, S. C., “Plant density studies with direct seeded tomatoes in Ontario, Canada,” Scientia horticulturae, Vol. 1 No. 4, p. 309-320, 1973 17 Stoelzle, A., “Simulation and Optimization of a Greenhouse for long term Space Missions,” Semester Thesis, Technical University Munich, RT-SA-2013/17, 2013. 18 Kirschner, F., “Mass and Energy Balance Models for Rodents in Space Life Support Systems,” Bacherlors Thesis, Technical University Munich, RT-BA-2013/19, 2013. 19 Olthoff, C., “V-SUIT An Approach Towards a Virtual Space Suit,” AIAA 2012-3608, 42nd ICES, 2012, San Diego. 20 Portner, B., “Numerical Modeling of a Fixed Bed Adsorber for Carbon Dioxide Removal in Human Space Flight,” Bachelors Thesis , Technical University Munich, RT-BA 2013/15, 2013. 21 Stegmiller, M., “Simulation and Testing of Amine-Based Vacuum Swing Adsorption Units for Carbon Dioxide and Humidity Control,” Bachelors Thesis , Technical University Munich, RT-BA 2013/08, 2013. 22 P¨ utz, D., “Dynamic Simulation Model of Heat Exchangers,” Bachelors Thesis , Technical University Munich, RT-BA 2013/10, 2013. 23 Rector, T. J., Steele, J. W., and Bue, G., “Performance of a Water Recirculation Loop Maintenance Device and Process for the Advanced Spacesuit Water Membrane Evaporator,” Proceedings of the 43rd International Conference on Environmental Systems, AIAA 2013-3308, Vail, Colorado, USA, 2013, doi: 10.2514/6.2013-3308 24 Makinen, J. V., Anchondo, I., Bue, G., Campbell, C., and Colunga, A., “Next-Generation Evaporative Cooling Systems for the Advanced Extravehicular Mobility Unit Portable Life Support System,” AIAA 2013-3340, Proceedings of the 43rd International Conference on Environmental Systems, Vail, Colorado, USA, 2013, doi: 10.2514/6.2013-3340 25 Bue, G. C., Trevino, L., Tsioulos, G., and Hanford, A., “Testing of Commercial Hollow Fiber Membranes for Spacesuit Water Membrane Evaporator,” 2009-01-2427, Proceedings of the 39th International Conference On Environmental Systems, Savannah, Georgia, USA, 2009, doi: 10.4271/2009-01-2427 26 Braun, J., “Simulation of the Space Suit Water Membrane Evaporator in V-HAB,” Bachelor’s Thesis, LRT-BA 2013/05, Technical University Munich, Institute of Astronautics, Munich, 2013. 27 Bue, G., Trevino, L., Tsioulos, G., Settles, J., Colunga, A., et al., “Hollow Fiber Spacesuit Water Membrane Evaporator Development and Testing for Advanced Spacesuits,” AIAA 2010-6040, Proceedings of the 40th International Conference on Environmental Systems, Barcelona, Spain, 2010, doi: 10.2514/6.2010-6040 28 G¨ oser, J., “Development of a dynamic simulation model of a Liquid Cooling Garment,” Bachelors Thesis , Technical University Munich, RT-BA 2013/06, 2013. 29 Weber, T., “Extension and correlation of a dynamic simulation model of the Life Support System of the International Space Station,” Bachelors Thesis , Technical University Munich, RT-BA 2013/09, 2013. 30 Gruschke, F., “Reliability Analysis of physicochemical Life Support Systems for Long Duration Missions,” Bachelors Thesis , Technical University Munich, RT-DA 2013/06, 2013. 2 Jones,
22 International Conference on Environmental Systems