( ' •1 (NASA-CR-199125) MULTIOBJECTIVE N95-32688

, OPTIMIZATION OF HYBRID REGENERATIVELIFE SUPPORT TECHNOLOGIES. TOPIC D:TECHNOLOGY ASSESSMENT Progress unciasReport (California Univ.) 27 p

G3/54 0063071

NASA-CR-199125

NASA Interim Progress Report

May 1995

Project Title

Principal Investigator

Multiobjective Optimization of Hybrid Regenerative LifeSupport Technologies (Topic D: Technology Assessment)Vasilios Manousiouthakis, Associate Professor5531-J Boelter HallDepartment of Chemical EngineeringUniversity of California, Los AngelesLos Angeles, CA 90095(310)825-9385

wi

rlntro duct ion

'In the first year of this project we developed simple mathematical models for many of the technolo-gies constituting the water reclamation system in a space station. These models were employed forsubsystem optimization and for the evaluation of the performance of individual water reclamationtechnologies, by quantifying their operational "cost" as a linear function of weight, volume, andpower consumption. Then we performed preliminary investigations on the performance improve-ments attainable by simple hybrid systems involving parallel combinations of technologies.

In the second year of the project, we are developing a software tool for synthesizing a hybridwater recovery system (WRS) for long term space missions. As conceptual framework, we areemploying the state space approach [1]. Given a number of available technologies and the missionspecifications, the state space approach would help design flowsheets featuring optimal processconfigurations, including those that feature stream connections in parallel, series, or recycles.

We visualize this software tool to function as follows: given the mission duration, the crew size,water quality specifications, and the cost coefficients, the software will synthesize a water recoverysystem for the space station. It should require minimal user intervention. It-s=overalt=str-uctur.ejs~shown-in-Figure-l. The following tasks need to be solved for achieving this goal:

•'^Formulate a problem statement that will be used to evaluate the advantages of a hybrid WRSover a single technology WRS.'

••^Model several WRS technologies that can be employed in the space station.

/•^Propose a recycling network design methodology. Since the WRS synthesis task is a recyclingnetwork design problem, it is essential to employ a systematic method in synthesizing thisnetwork.

1

\.,- J

Mission Specifications WRS Design

Input Interface Output Interface

Optimization Routines

Heat Pinch Tool ASPEN Models Proprietary Models

WRS SYNTHESIS SOFTWARE

Figure 1: Structure of the proposed water reclamation system design software.

VDevelop a software implementation for this design methodology. Design a hybrid systemusing this software and compare the resulting WRS with a base-case WRS. I |

t /"•

Create a user-friendly interface for this software tool.

* '-̂ ^̂ -sf̂ r

Ourjsffor-ts-towards-addressingjthese tasks are-0u-t-Mned*freiowK^fc,-,

2 Problem Definition

The associated recycling network design task is:

Given a set of feed streams that need to be recycled, with known flowrates, andinlet compositions and enthalpies; a set of product streams with known specificationson outlet compositions and enthalpies; a set of water reclamation technologies to beconsidered; and a set of auxiliary feed streams (e.g., air); determine the minimum"cost" system that delivers the product specifications.

The "cost" is a known linear function of weight, volume, and power consumption.This problem statement is shown in Figure 2. The solution obtained to the above optimization

task is the required WRS configuration that is indicated as a box in the figure.

Feeds Products

Urine & Flush

Shower Water

Cloth Wash Water

Dish Wash Water

-*• Potable Water

•*• Hygiene Water

-*• Waste Water

-"• Cone. Urine

Figure 2: Input-Output diagram for a water recovery system. The physical properties of eachstream are shown in Tables 1 and 2.

2.1 Technical Specifications

To proceed towards solution of this recycling network design task, we must first determine the inletconditions of our feed streams and the outlet specifications on our product streams.

As feed streams, we consider four water streams: urine and flush, shower water, cloth washwater, dish wash water; their inlet compositions are given in Table 1. In reality, the space stationwastewaters contain over 150 components. Most of these components are present in trace amounts.However, in Table 1, we neglect most of these trace components, and focus primarily on the com-ponents considered in our previous report [5]. The exact inlet compositions for each component arebased on the data in [4], but some chemical species have been consolidated into one representativechemical of the same category. Though there are other waste water streams in the space mission,only the above four have significant flowrates to be reusable.

The water recovery system must deliver two purified water streams with different specifications:potable water and hygiene water. Specifications on these streams are shown in Table 2. Thesespecifications have been derived from [3]. There are two waste effluents from the purificationprocess: effluent water and concentrated urine. Part of the effluent water stream could be recycledback into the process.

In the entire process, we have restricted the water streams to contain only 11 components. Thissimplified representation yields optimization problems that are easier to solve. The componentsconsidered in the streams are: urea, creatinine, r^O, NaCl, phenol, soap, and NHs. Besides thesespecies, 62, N2, N20, CC>2 are also considered, since they are present in the two reactors of theVPCAR subsystem.

The heat loss or gain to the environment from the system is modeled by considering the envi-ronment air as the heating or cooling medium [4].

The above feed and product streams constitute the typical input-output specifications for awater recovery system. The proposed software tool will use these specifications. However, the usercan alter them to match the needs of a particular mission.

Temp(K)Pres(Pa)Flow Rate(kg/day)UreaCreatinineH20NaClPhenolSoap

NH3

C02

02

N2

N20

Urine &Flush

297.2694400.00

1.8962E-31.0024E-25.0651E-39.7717E-17.2132E-35.2698E-4

0.00.00.00.00.00.0

ShowerWater

360.002.0137E5

6.1818E-49.2045E-41.7008E-39.9099E-16.7813E-4

0.05.7031E-36.4706E-6

0.00.00.00.0

Cloth WashWater

360.002.0137E5

1.3763E-36.9047E-41.3009E-39.9611E-14.7736E-4

0.01.4010E-31.9981E-5

0.00.00.00.0

Dish WashWater

360.001.0137E5

6.2847E-40.00.0

9.9490E-10.00.0

5.1000E-30.00.00.00.00.0

Table 1: Feed stream properties. The composition of each species is given as its mass fraction.

Mass Frac. Hygiene Water | Potable Water

Temp(K)Pres(Pa)

Urine and CreatinineNaCl

PhenolSoap

NH3

C02

02

N2

N20

297.262.0137E51.100E-53.751E-4l.OOOE-95.0E-75.0E-71.5E-5l.OE-5l.OE-5l.OE-6

297.261.0137E56.000E-73.751E-4l.OOOE-95.0E-75.0E-71.5E-51.5E-51.5E-51.5E-6

Table 2: Water quality requirements adapted from Carrasquillo (1992), stated as maximum allow-able impurity mass fractions.

3 WRS Technologies: Model Development

Several technologies have been developed to recycle the wastewater streams into potable and hy-giene streams detailed above. These include: thermoelectric integrated membrane evaporationsubsystem (TIMES), vapor compression distillation (VCD) subsystem, vapor phase catalytic am-monia removal (VPCAR) subsystem, and the closed cycle air evaporator(AE) subsystem. Theproposed WRS synthesis tool will employ these four technologies as constituents of the recyclingnetwork. Additional technologies can also be incorporated if desired.

For TIMES and AE, modeling equations detailed in our previous report [5, TM8-15 and AE13-15] are employed. For VCD and VPCAR, we have developed a simplified ASPEN model (explainedbelow) which is based on [4]. These models quantify the operational "cost" of each technology asa linear function of weight, volume, and power consumption.

4 Recycling Network Design Through The State Space Approach

Solution to the recycling task stated above will yield the optimal configuration of technologies tobe employed in the space station WRS. This configuration is selected as optimal among severalconfigurations captured by the state-space description of the WRS.

As an example of the state space representation for the recycle design task, consider one possibleconfiguration for a WRS shown in Figure 3. It employs two technologies: vapor compressiondistillation (VCD) and vapor phase catalytic ammonia reduction (VPCAR). It employs four vapor-liquid separators, two reactors and several pumps and heat exchangers. There are four feed streamsand four product streams, as stated above.

The state space representation for this WRS is shown in Figure 4, and consists of a distributionnetwork (DN), a heat exchange operator (HEO), and a process operator (PO). This state spaceapproach can be employed to represent any flow configuration in this water recovery system. Anyfeed stream that enters the DN can leave the DN either as a product stream or as a stream enteringthe heat exchange operator. Streams leaving the two operators return to the DN. These, in turn,either exit the DN as a product stream or mix with the streams entering the heat exchange operator.Thus, the DN consists of inlet or splitting junctions, exit or mixing junctions, and all possiblestreams connecting these junctions. For example, a stream (LEXP) from the HEO returns to theDN, where it splits into a product stream (BTM) and a recycle stream (VXR2) to the HEO. Ina general DN, this stream can split into as many streams as there are mixing (exit) junctions inthe DN. Within the DN, each mixing or splitting junction must satisfy the overall and componentmass balances, and the energy balance.

Pressure changes, namely, compression and expansion, are also represented within the DN.For example, the vapor product (50) from the flash separator in the vapor compression distilla-tion (VCD) subsystem, may be compressed and heat a feed (VXEV) for a reactor (RXR1) inthe VPCAR. However, such pressure changes increase the power consumption of the system, andcontribute to the objective function. Note that mixture streams can be neither compressed norexpanded.

Figure 3: A water recovery subsystem configuration employing vapor compression distillation(VCD) and vapor phase catalytic ammonia reduction (VPCAR) technologies.

Figure 4: The state space representation of a water recovery subsystem employing VCD and VP-CAR.

All streams leaving the DN, except the product streams, enter the heat exchange operator. Inthe heat exchange operator, the streams are either heated or cooled to the desired exit temperatures.The minimum hot and cold utility required to accomplish this task is calculated by employing thepinch methodology. Given a set of streams that need to be either heated or cooled, a set of hotand cold utilities, and a driving force for heat transfer, the HEO identifies the minimum utilityconsumption. The calculation of this minimum utility is based on the satisfaction of the first andsecond laws of thermodynamics. This calculation procedure has been automated in our "pinch"calculation tool.

Streams leaving the heat exchange operator either return to the DN or enter the process oper-ator. The streams entering the PO undergo a transformation due to unit operations of the severalavailable water reclamation technologies (TIMES, AE, VCD, or VPCAR). The unit operations aremodeled in simulator routines that employ either our own models, or those in commercial simulatorssuch as ASPEN. Finally the streams return to the DN as inlet streams.

Note that any stream entering the DN can split into several stream, each of which can mixwith a stream leaving the DN. Thus, while the DN can represent all possible flow configurationsbetween these splitting and mixing junctions, it creates an overwhelming decision task for the systemdesigner. Instead, the state space formulation, leads to an. optimization program to determine theflow configuration that achieves the product specifications at the least cost, from these options.

5 State Space Optimization Software

To solve this WRS design problem, the resulting optimization problem will be embedded in thesoftware tool. This requires the automatic generation of constraints that define the DN, the HEOand the PO; the simulation routines for each technology to be employed once they are specified;and the objective function, given the cost coefficients.

A nonlinear optimization package is employed for the solution of the resluting optimizationproblem. We have written FORTRAN subprograms that define the constraints for the DN, theHEO, and the cost function. We are also developing ASPEN-Plus simulation models for some of thetechnologies. An interface that combines the ASPEN-Plus unit operation models with the nonlinearoptimizer has been developed. This interface, for example, given the inlet and operating conditionsfor a flash separation in the VCD by the optimization routine, simulates the flash through ASPEN,and then returns the calculated flash outlet conditions to the optimization routine.

We have also acquired and installed new versions of ASPEN-Plus (Release 9.1.3) to help developthe software tool. This installation led to compatibility problems in running the IWRS flowsheetprovided in [4].

As a starting point for the optimization program, we have decided to simulate the IWRS-1 flowsheet [4] with only two technologies: VCD and VPCAR. Our simulation results for thisflowsheet are detailed in Appendix A, and the problems encountered in this simulation are outlinedbelow.

6 Problems

• We assumed that the dishwashing wastewater is composed of water and soap only. Thisfollows a similar assumption in [4]. However, this wastewater should have other organiccomponents, primarily oils and food debris. We would like to obtain a detailed compositionof this stream compatible with the 11 component representation we currently employ.

• We are missing accurate data for the feed and product compositions. Though the developmentof our simulation method does not depend on the exact inlet/outlet composition specifications,the results obtained from the program do. Hence, we would like at least one set of complete,representative stream data for a typical water recovery system. These specifications shouldcontain:

— The detailed inlet compositions with respect to the component list we have considered.

~ The outlet specifications (MCL) for the same list of components, that is based on therequirements in [3].

- Representative temperature and pressure conditions for the inlet and outlet streams.

• Base case simulation. In order to determine the detailed product specification table, wesimulated ASPEN flowsheets available from NASA Ames Research Center [4]. The results ofthis simulation are attached in Appendix A. The resulting product compositions are shownin Table 3.

However, these simulation results do not meet the specifications in Table 2. We would like toreconcile our flowsheet and simulation with these specifications.

7 Future Plans

In the work thus far on this project, we have made significant progress towards formulating andprogramming the optimization problem for the synthesis of a WRS. In the remaining project period,we have scheduled the following activities:

• Water quality specifications, given in Table 2, have been adapted from [3]. For instance, ureaand creatinine have been combined to meet the maximum allowable total organic content limitin the product %vater streams. A literature search for more detailed water quality requirementswill be pursued.

• Develop a MINOS-ASPEN-Plus interface to determine the stream thermodynamics for cal-culating the temperature, pressure, or enthalpy conditions at any junction in the state spacerepresentation.

• Improve the ASPEN-Plus model for the VCD and VPCAR subsystems.

• Combine the modules of the code to create the WRS synthesis tool.

• Perform a water reclamation system synthesis case study.

8 Conference Presentations

Our paper titled "Multiobjective Optimization of Hybrid Regenerative Life Support Technologies,"has been accepted for the 25th International Conference Environmental Systems. It will be pre-sented in July, 1995. A draft of this paper is attached.

10

References

[1] Bagajewicz, M. and V. Manousiouthakis (1992), Mass/Heat-Exchange Network Representationof Distillation Networks. AIChE Journal, 38, 1769-1800.

[2] Ballin, M. and Bilardo, V. J., Initial Lunar Output Life Support System Water Recovery Syn-thesis Results.

[3] Carrasquillo, R. L., D. L. Carter, D. W. Holder, C. F. McGriff, and K. Y. Ogle (1992), SpaceStation Freedom Environmental Control and Life Support System Regenerative Subsystem Selec-tion. NASA Technical Memorandum 4340. George C. Marshall Space Flight Center, Alabama.

[4] Flynn, M., J. Fisher, and M. Highttower (1991), Evaluation of a Potential Regenerative WaterReclamation Subsystem Design Using the ASPEN PLUS Simulation Program. NASA AmesResearch Center, Moffett Field, CA.

[5] Manousiouthakis, V. (1994), Multiobjective Optimization of Hybrid Regenerative Life SupportTechnologies. Department of Chemical Engineering, University of California-Los Angeles, CA.

[6] Putnam, D. F. (1970), Composition and Concentrative Properties of Human Urine NASA Con-tract # NAS1-8954, McDonnell Douglas Aeronautics Company, Huntington Beach, CA.

[7] Wydeven, T and M. A. Golub (1991), Waste Streams in a Crewed Space Habitat. Waste Man-agement & Research, 9, 91-101.

[8] Wydeven, T. (1983), Composition and Analysis of a Model Waste for a CELSS. NASA TechnicalMemorandum 84368. NASA Ames Research Center, Moffett Field, CA.

[9] Wydeven, T. and Golub, M. (1990), Generation Rates and Chemical Compositions of WasteStreams in a Typical Crewed Space Habitat. NASA Technical Memorandum 102799. NASAAmes Research Center, Moffett Field, CA.

11

A ASPEN-PLUS Simulation

ASPEN-PLUS is a commercial chemical process simulator. We have available to us the ASPEN-Plussimulation for one possible water recovery system configuration. This simulation was developedat the NASA Ames research center as part of a life support subsystem evaluation tool [4]. Theflowsheet for the water recovery system consists of three subsystems: VCD, VPCAR, and super-critical water oxidation (SCWO). The NASA-Ames code simulates the separation tasks in eachof these technologies with a combination of flash columns, heat exchangers, mixers and splitters.This approach allows one to simulate the novel technologies of IWRS on a commercial processsimulator. The NASA-Ames model involves over 30 components or chemical species, includingtrace electrolytes and solids, in the flowsheet. Several of these components for which detailed ther-modynamic properties are unavailable, were modeled with the same thermodynamics as those ofwater.

Even though human urine is composed of over 150 constituents, this model simulated 26 compo-nents for urine. These reduced components were first proposed in [6]. The compositions of showerwastewater, cloth washing wastewater and dishwashing wastewater are based on [9]. The flow ratesof these streams are adapted from [2].

The WRS flowsheet has been simulated on a new release of ASPEN-Plus. Unfortunately, theNASA-Ames code, written for an earlier release of ASPEN, is not compatible with this release.Therefore, we have recreated an ASPEN model for VCD and VPCAR technologies, to simulateWRS configurations derived from the state space program. This simulator serves two other pur-poses: it provides feasible starting points for the optimization program, and parts of the simulatorcan be employed in the state space optimization.

The simulation of the VCD/VPCAR flowsheet involved several modifications in the originalsimulator. First, we extracted VCD and VPCAR information from the NASA-Ames flowsheet.Then, the following changes were made:

• The electrolyte and solid components were neglected to simplify the problem.

• We reduced the 30+ components in this flowsheet to 11, to make our optimization problemmore tractable.

• Three flash columns in the VCD were deleted because they were adjusting electrolyte equi-librium temperature differences, that were no longer required.

• FORTRAN program units that controlled flow rates in the splitters were deleted.

• A reactor employed for urea hydrolysis reaction was removed because it produced low con-versions, and led to convergence errors during program execution.

• The area of a heat exchanger in the VCD was increased to make it feasible.

• Feed 0% required for an oxidation reaction was deleted. Instead, it was assumed that enough02 for this reaction is always available.

12

Temp(K)Pres(Pa)Flow Rate(kg/day)

UreaCreatinineH2ONaClPhenolSoapNH3

C02

02

N2

N20

PotableWater

373.19501.0134E57.6926E-4

2.9433E-105.5045E-109.9999E-1

0.01.1055E-68.0179E-85.4971E-91.0074E-6

0.03.6817E-94.8429E-8

HygieneWater

373.19501.0134E52.3078E-3

2.9433E-105.5045E-109.9999E-1

0.01.1055E-68.0179E-85.4971E-91.0074E-6

0.03.6817E-94.8429E-8

WasteWater

300.00001.0134E51.8324-03

0.00.0

9.9970E-10.0

6.4957E-80.0

3.3459E-72.6246E-43.9272E-61.3798E-5

0.0

Cone.Urine

345.568394400.001.0010E-5

4.1504E-27.7545E-26.2037E-12.3742-026.0838E-4

0.23625.3826E-8

0.00.00.00.0

Table 3: Product stream properties from the ASPEN-Plus simulation of a VCD-VPCAR configu-ration. The composition of each species is stated as its mass fraction.

• One of two consecutive separators in VPCAR was removed to improve flowsheet convergence.

This modified flowsheet is illustrated in Figure 3 and the simulation results are shown in Table 3.It should be noted that the output compositions of these streams does not satisfy all the waterquality requirements. We are currently investigating flowsheet modifications that will allow allwater quality requirements to be met.

13

6! 6 j

si:

8 8

§116| 9

8 8 8d| d 9

Mil

si?

9

O! <

8 8

I

I

M I CM0 |Z

8 8O 0

8! 18d: id

818d i d

8 Ei! <

Multiobjective Optimization of Hybrid

Regenerative Life Support Technologies

V. Manousiouthakis, D. Sourlas, M. Wilcoxson, J. Choi, S. Han, and H. Kim

Department of Chemical EngineeringUniversity of California, Los Angeles

Los Angeles, CA 90095

ABSTRACT

This paper presents a methodology to design op-timal water reclamation systems for long-durationspace missions. These systems employ, separately,numerous technologies and are required to satisfymultiple goals. However, each of the available tech-nologies may not meet these goals by itself. In-stead, it is proposed to evaluate the optimal designof such systems employing, simultaneously, severaltechnologies. First, mathematical models for eachtechnology considered are developed. These mod-els are employed to optimize two of these technolo-gies, demonstrating that significant "cost" reduc-tions are possible. Finally, hybrid systems withmultiple technologies are designed and optimizedto meet multiple objectives. The obtained resultsshow that these systems outperform single technol-ogy based water reclamation systems.

INTRODUCTION

To reduce payload requirements that wouldmake long-term space missions feasible, a closed-loop life support system for the mission crew is es-sential. This environmental control and life sup-port system (ECLSS) is comprised of water recla-mation, air revitalization, and waste management.Since water needs constitute 94% of the payload ofthe expendables [1], research has been focussed onwater reclamation.

For a long-duration mission, water is requiredfor drinking and food preparation for the crew. It isneeded for hygiene purposes: shower, laundry, flushand dishwashing. Laboratory experiments that thecrew carry out during a mission also may require

water for cleaning and reagent use. Water is ex-creted in the sweat, urine, and breath of the crew.Water vapor is also produced from the hygiene sys-tems. The purity requirements for potable, hygieneand laboratory water are different. Similarly, thecomposition of water from these sources varies. Theaim of the water management unit is to convert theexcreted or effluent water streams into streams thatmeet the quality specifications for drinking, hygiene,or laboratory. The water can then be recycled.

The different water requirements pose the prob-lem of designing a water reclamation system thataccomplishes the recycling of the different waste wa-ter streams. This problem can be stated as follows:

Given a space flight mission, the spec-ifications on the potable, hygiene andmiscellaneous water streams, and thecompositions of the waste streams fromthese functions, synthesize a network ofseparation processes, material-handlingunits and energy-recovery devices thatremoves, in a micro-gravity environ-ment, the undesirables and convertseach of the waste streams to recyclableproduct streams with maximum relia-bility and safety, and minimal volume,mass, new expendables and power re-quirements.

One technology alone cannot efficiently handle theserequirements. For instance, if flash distillation is thechief unit operation, adsorption and ion-exchangemay be a part of the system for pre/post-treatment.Given these multi-objectives and multi-technologyoptions for the ECLSS, a method is required thatwould evaluate and choose the optimum technology

combination and design a system that meets thevarious objectives at hand.

Ongoing research in this area has led to thedevelopment of several devices and regenerationschemes that utilize them. Next, we briefly reviewthese technologies.

WATER RECLAMATION TECHNOLOGIES

We consider six subsystems: thermoelec-tric integrated membrane evaporation subsystem(TIMES), multifiltration (MF), vapor compressiondistillation (VCD), vapor phase catalytic ammoniaremoval (VPCAR), reverse osmosis (RO), and airevaporation (AE). Below, these subsystems are dis-cussed briefly.

TIMES - In this process [3, 8, 20, 21], contam-inated water is heated and evaporated in a hollowfiber membrane (HFM) evaporator which functionsas a pervaporation unit. In this unit, water diffusesthrough the membrane and evaporates as soon asit reaches the outside surface of the tubes which iskept at low pressure by means of a vacuum pump.The vapor from the pervaporation unit is condensedon a porous plate connected to the cold junctionsof the thermoelectric elements and its latent heatis recovered through a thermoelectric heat pump.This heat, combined with the Joule heat generatedbecause of the inefficiency of the thermoelectric ele-ments, is used in the heater that precedes the HFMevaporator.

MF - In this subsystem [15, 16], contaminatedwater is purified by directing it through a series ofcolumns, each consisting of several different adsor-bents. Each bed material adsorbs different com-ponents of the feed stream. This is a reliable andsimple technology which requires no micro-gravityphase separation since the solid is stationary. Itcould serve as an extremely competitive techniquein tandem with other technologies that require somepre- or post-treatment of the waste feed.

VCD - Vapor compression distillation is a phase-change separation procedure designed to recoverwater from its solutions (such as urine). The prod-uct from this process is hygiene water that can besubsequently processed by a multifiltration or a re-verse osmosis unit. An evaporator and a mechani-cal compressor constitute the main components of aVCD system. The vapor produced in the evaporatoris compressed so that its latent heat vaporizes thewaste water stream entering the evaporator. In this

system effective heat integration has been employedto reduce the overall power requirements. However,when VCD is considered as part of a water reclama-tion system that involves several technologies, it isnot obvious whether this type of integration remainsoptimal.

VPCAR - The wastewater stream is va-porized in a hollow fiber evaporator and thentreated to remove ammonia and light hydrocar-bons. The treatment is a two step catalytic oxida-tion/decomposition of these compounds. The chiefadvantages of this process is that it does not re-quire post and pre-treatment of water, delivers highquality water, and involves a small number of mov-ing parts. However it requires large specific energybecause of the high temperature reactions.

RO - Reverse osmosis (hyperfiltration) has beenstudied for wash water purification. In this process,an applied pressure counters the osmotic pressureto force flow across the membrane from the highsolute concentration side to the low solute concen-tration side. The concentration difference across themembrane arises from the preferential permeabilityof water compared to contaminants (solutes) suchas salt, soap, and urea. The hollow fiber RO mod-ule minimizes weight and maximizes resistance tofouling [17].

AE - Warm dry air is blown through a wick toevaporate the contaminated water. The warm moistair leaving the evaporator is then cooled to recoverthe water. The cooling takes place in two steps.The first heat exchanger (recuperator) is includedto reduce the duty for both the heater and the con-denser. The cold saturated air from the condenser issent through the recuperator and then the heater toregenerate warm dry air. It would be more efficientto have multiple wick stages, rather than one stageas in the tested configuration. A multistage designwould also require new heat integration schemes.

The optimization and technology assessment ofa water reclamation system requires simple and re-liable mathematical models for each subsystem. Wehave developed such models for the above subsys-tems. These models are derived from first princi-ples and are sufficiently detailed to allow predictionof several quantities of interest (e.g. power, weightand volume) while still remaining tractable for usein large scale optimization studies. These modelsand equations are described in [9].

Next, these models are employed to optimizetwo subsystems.

SUBSYSTEM OPTIMIZATION

Subsystem performance can be optimized byappropriate choice of operating condition values.Next, we illustrate this approach using two par-ticular subsystems (wick evaporator and vaporcompression distillation) and we demonstrate howoptimization can lead to performance improve-ments. In order to evaluate a system's performance,one should consider various characteristics such asweight, power consumption, volume, etc. In ourwork, we employ a single performance measure thatsimultaneously takes into account these contribu-tions and quantifies the system's performance interms of the total equivalent of the system as:

Table 1: Wick Comparision

Uj = CpPTT (1)

All symbols are explained in the notation sectionat the end of the paper and cost coefficients Cw,Cp, and Cv are taken from [19]. Then, (1) becomesthe objective function in a minimization problemthat is solved using standard optimization software,for example MINOS. Since (1) emcompasses variouscontributions to the cost of the system, the pro-posed optimization approach is inherently multiob-jective. Overall process model was used to relateoperating conditions to weight, power and volumerequirements. These equations were constraints inthe optimization problem.

AE SUBSYSTEM OPTIMIZATION - Air evap-oration water recovery has proven successful forurine processing [11, 13, 14]. The design of sucha system requires that operating condition such asair flow rate, condenser temperature and wick inlettemperature be specified. In our work these valuesare such that the equivalent cost of the AE systemis minimized.

The system was assumed to process 10 kg/dayof urine flush (3-man crew) and the mission timewas specified as 1 year. The results are tabulatedin Table 1. For Case 1 condensation is permittedin the recuperative heat exchanger, while for Case2 no phase change is allowed. The optimal designsfor Case 1 and Case 2 differ in their temperaturesand flow rates (the condenser temperature changesfrom 10°C to 28.3°C), but the objective function isnearly unchanged. Either optimal design resulted ina decrease in the objective function of nearly 30%as compared to the original design proposed in [11].This reduction was due to the reduced heating andcooling requirements (Qn and Qc) in the heater, re-

Variable*F(mole/s)xwo•X-wiV•"•wot

fJJ / /"t\

•^ tut I ̂ ^ /FT! / f^\

Tl"(C)TWOf(C)Td(C)L(cm)Tfi(C)Xho

Qc(W)Q^(W)/,*,-($ x 106)

Morasko et. al.0.49

0.0250.0120.05254.410.033.921.122.840.021.1

0.0254634351.78

Case 10.41

0.0280.0120.08665.010.043.723.020.523.338.0

0.0243212881.29

Case 20.51

0.0510.0380.11065.028.347.433.333.321.842.4

0.0513352941.33

* all symbols explained in the notation section.

cuperator and condenser, and the reduction in thevolume and mass of the air evaporator due to a de-crease in its length (L).

VCD SUBSYSTEM OPTIMIZATION - Theperformance of the VCD unit for different valuesof the pressure ratio across its compressor (Pz/Pi)is quantified. Again, the system's performance ismeasured through the, so called, equivalent cost ofthe system (1). In this way, the different contri-butions to the cost of a mission are appropriatelyweighted and their sum yields the equivalent cost.

Three contributions to the cost of the missionwill be considered in this study of VCD: weight (ofrotating drums and their cylindrical covers), volume(due to the cylindrical covers) and work (compres-sion work). The corresponding cost coefficients arethe same as the ones used in (1). Changes in thepressure ratio do not significantly affect devices inthe VCD flowsheet, other than the rotating drumassembly. The contribution of such devices to theoverall cost can be assumed to be constant, and willnot be considered in the subsequent calculations. Inthis way, optimal pressure ratio values can be identi-fied, and the sensitivity of the cost value to changesin the pressure ratio can be established..

The following additional system parameters arefixed in this calculation [12]:

Equivalent Cost vs Pressure Ratio

Equivalent Cost vs Pressure Ratio(Pi = 35torr)

io3 3

140.0

120.0

100.0

80.0

60.0

40.0

i i 1

PI = 30 torr OPI = 35 torr +Pi = 40 torr DPi = 45 torr X

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0Pressure Ratio Across Compressor

Equivalent Cost vs Approach Temperature

10s $

140.0

120.0

100.0

80.0

60.0

40.0

i i iPI = 30 torr OPI = 35 torr +PI = 40 torr DPI = 45 torr X

5.0 10.0 15.0 20.0 25.0 30.0 35.0

Approach Temperature (K)

Figure 1: Optimal Pressure Ratio and Approach Tem-perature for VCD drum

Material: Aluminum (density 2.64 g/cm3)Metal Thickness: 1 mmSmall side of Inner Drum: radius 20 cmTaper Angle: 5°Clearances: 0.2 times the radius of the nearestdrum side.

The results of this study are summarized in Fig-ures 1 and 2. Low pressure ratios lead to smallenergy consumption at the compressor. They alsolead to small differences in evaporation tempera-tures, i.e. small approach temperature values atthe VCD drum. In this case, to achieve the desiredheat exchange, large drum areas are required. Asa result, volume and weight are expected to be thedominant contributions in the equivalent cost. Thisis supported by Figure 1 which establishes that op-erating a VCD unit with a pressure ratio value closeto 1 is not an optimal policy.

The curves in Figure 1 reach a minimum whichis located outside the pressure ratio range shown inthis figure. The location of this minimum depends

10s

1400.0

1200.0

1000.0

, 800.0

600.0

400.0

200.0n n

A 1

j-i

<2>X

L x+n Xu

_LE

"o+S" °<*>5

i i i i

1 month O

10 months +

20 months D

50 months XXXXXXXXXXXXXXXXXXXXXXX>G

'rttQ 1 1 1 1 | | | 1 1 1 | I 1 1 1 1 1 1 L

^^V\AAy\XiXXXXXjXXXXXDOOOOC

1.0 1.5 2.0 2.5 3.0 3.5

Pressure Ratio Across Compressor

4.0

Figure 2: Equivalent cost for PI = 35 torr and variousmissions.

on the mission specifications. Figure 2 shows theequivalent cost as a function of the pressure ratiofor various mission duration. As the length of themission increases, the minimum of the equivalentcost curve appears at smaller pressure ratio values.Furthermore, the equivalent cost curves that corre-spond to longer missions possess more pronouncedminima. Thus, as the mission duration increases,the range of pressure ratios that can deliver per-formance close to the optimum becomes narrower.From Figure 2 it can be concluded that a pressureratio value of approximately 2.5 delivers acceptablesystem performance for all missions in the rangeconsidered.

The above results show that employing an opti-mization based approach to the design of an inde-pendent subsystem, based on mathematical modelsand an equivalent cost objective function, can yieldsignificant cost reductions and physical insights. Inthe sequel, the design of an integrated water recla-mation system is pursued.

SYSTEM INTEGRATION

Typical systems that recover hygiene water fromurine flush involve a separation process (evapora-tion, pervaporation or air evaporation) followed bya multifiltration step for product purification. Typ-ically the subsystems involved in this process aredesigned and optimized independently from eachother. However, even if the individual subsystemsare optimized, there are no guarantees that the re-sulting overall system will also be optimal. The

overall system is first represented in terms of allpossible interconnections among the allowable tech-nologies/subsystem, and then it is optimized. Inthis way, all processing technologies and their inter-action are simultaneously considered.

Two avenues for the synthesis of optimal waterrecovery systems were pursued: heat integration,and multiobjective hybrid system optimization.

HEAT INTEGRATION - Given a set of hot andcold streams of a process, heat integration tech-niques, in particular pinch analysis, identify the in-herent thermodynamic limitations for heat transferof this process. In addition, the minimum hot andcold utility necessary to achieve the thermodynamicspecifications for the process streams is also identi-fied. The resulting process flowsheets feature ex-change of heat between the hot and cold streamsso that the external energy input to the process isminimized. Applying this approach to water recov-ery processes will help identify energy efficient pro-cesses.

Consider a typical water reclamation system forurine flush employing a two step separation. Thefirst step consists of VCD, or TIMES, or a combi-nation thereof (i.e. hybrid system). In the secondstep, the distillate from the first step is processedthrough a multifiltration unit. The operating con-ditions assumed for this analysis are the following:

Urine Flush: 1 kg/hrEvaporation Temperature (VCD): 31.5°CEvaporation Temperature (TIMES): 59°CMinimum Approach Temperature (ATmjn) : 5°CSupply & Target Temperature for all streams: 20°C

The inlet stream to the MF unit is first sterilizedthrough heating to 82°C and then cooled to a fi-nal temperature of 20°C before passing through theunibed. The target temperature for all streamsleaving this process is 20°C, equal to the tempera-ture of the inlet stream. The system configurationsthat were examined are the following:

SI: VCD unit (no compressor) followed by MFunit (with heat integration).

S2: TIMES unit (no heat pump) followed by MFunit (with heat integration).

S3: 50% of the feed through VCD (as in SI) and50% of the feed through TIMES (as in S2)followed by MF unit (with heat integration).

S4: VCD unit followed by MF unit (with compres-sor, latent heat recovery, no heat integration,independent design).

S5: TIMES unit followed by MF unit (with ther-moelectric heat pump, latent heat recovery,no heat integration, independent design).

S6: VCD unit followed by MF unit (with heat in-tegration).

S7: TIMES unit followed by MF unit (with heatintegration).

S8: 50% of the feed through VCD (as in S6) and50% of the feed through TIMES (as in S7)followed by a MF unit (with heat integration).

Pinch calculations were involved in the study of sys-tems Si to S3 and S6 to S8 since these are the onlysystems that are heat integrated. These calculationswere performed by means of an in-house developedsoftware program.

Cases SI, S2, and S3 correspond to various lev-els of system integration, whereby heat integrationand the potential of hybrid systems are explored.In these cases, the vapor compressor system inVCD and the thermoelectric heat pump systems inTIMES have been omitted, thus the required poweris zero for all systems in this category. Pinch calcu-lations were employed and the resulting minimumutility requirements are shown in Table 2.

Table 2: Minimum Utility Requirements

Case

SIS2S3

MinimumHeating

Utility(W)645830438

MinimumCooling

Utility(W)644615330

PowerRequirement

(W)000

Table 3: System Design with no Heat Integration

Case

S4S5

HeatingRequirement

(W)5621

CoolingRequirement

(W)6971

PowerRequirement

(W)16195

100

80

Temp 60

°C

40

20

I I

Hot Util.

Hot CUTT« Cold Curre

Cold Util.

I

0.0 1.0 2.0 3.0 4.0

Heat Load

Figure 3: Composite Curve Diagram corresponding toSI

We observe significant minimum heating andcooling utility requirements. Let us consider Fig-ure 3 which depicts the composite curve diagramfor system SI. Both the hot and the cold curve,in this diagram, have a plateau that correspondsto the condensation of the distillate and the evap-oration of water respectively. The size of theseplateaus suggests that phase changes are the dom-inant processes from an energy consumption pointof view. In the absense of equipment that upgradeheat (compressors, heat pumps etc.), recovery of thelatent heat content of the condensing vapor streamthrough heat exchange with the vaporizing stream isthermodynamically infeasible. To ensure thermody-namic feasibility for the overall process and achievethe process targets, the consumption of hot and coldutilities is required. From Table 2 it is establishedthat the hybrid system S3 requires less hot and coldutilities than SI or S2.

Cases S4 and S5 help quantify the impact thatequipment that upgrade heat have on the design ofwater reclamation systems that involve VCD andTIMES. Simple heat and mass balances lead to Ta-ble 4.

Configurations S4 and S5 feature latent heat re-covery. The use of the compressor (in VCD) andthe thermoelectric heat pump (in TIMES) upgradesthe heat contained in the condensing streams thusheat exchange with the vaporizing streams withineach process becomes feasible. Although the figuresshown in Table 4 do not correspond to minimumutility requirements, a significant reduction in util-ity consumption, as compared to cases Si and S2,is observed. The penalty is that electrical power is

consumed by the new system. By comparing S4 andS5 based on the information in Table 4, it is seenthat TIMES requires less hot utility than VCD. Thisis mainly attributed to the higher temperature ofits product stream compared to the correspondingstream from VCD (59°C vs 31°C). Similarly thecooling requirement is higher for TIMES becausethe brine is at a higher temperature (60°C vs 31°Cfor VCD).

To evaluate the potential of the proposed ap-proach for system integration, cases S6, S7 and S8were developed. Cases S6 and S7 are similar to S4and S5 with the addition of heat integration. CaseS8 corresponds to a 50% TIMES- 50% VCD hybridsystem.

Table 4: System Design with Integration

Case

S6S7S8

MinimumHeating

Utility (W)3.33.33.3

MinimumCooling

Utility (W)225337

PowerRequirement

(W)16195106

Figures 4, 5, and 6 present the composite curvesfor the systems S6, S7 and S8 respectively. Thedifferent scales in Figures 4, 5, and 6 represent thefact that although the heating and cooling require-ments for S6, S7 and S8 are comparable, the to-tal amount of heat transferred from the hot to coldstreams varies significantly among the three config-urations. Comparing Figures 3 and 4 one can easilyvisualize the effect of the heat upgrading capabilityof the compressor in VCD whereby the Hot Curvehas been sufficiently raised so that latent heat recov-ery is feasible (increase in the pressure of the vaporleads to increase of the condensation temperature).

The analysis performed in cases S6 and S7 leadsto subsystems similar to existing ones. This impliesthat the existing subsystem designs utilize heat andenergy efficiently. The instance of a hybrid systemconsidered in S8 suggests that from an energy ef-ficiency point of view, hybrid systems are not su-perior. This conclusion is subject to the restric-tions imposed in the definition of S8; namely thatfixed operating conditions for VCD and TIMES,and given subsystem designs be employed. Differentoperating conditions (i.e. different evaporation andcondensation temperatures etc.) can alter the shape

of the composite curves (Figure 5) and lead to flow-sheets with different characteristics. A parametricanalysis over a large temperature range requires in-formation regarding the quality of the product ofeither process as a function of its operating tem-perature. Furthermore the utilization of alternativeheat recovery techniques (heat pumping, produc-tion of electrical energy) within each subsystem alsoneed to be examined.

Temp

80

60

40

20

\ \ I I

Hot CUTTC

Cold CurTe

J I I I

-1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0

Heat Load 106 J/hr

Figure 4: Composite Curve Diagram corresponding toSB

Temp"C

80

60

40

20Cold Util.

-0.1 0.0 0.1 0.2

Heat Load 10* J/hr

0.3

Figure 5: Composite Curve Diagram corresponding toS7

MULTIOBJECTIVE HYBRID SYSTEM OP-TIMIZATION - Heat integration aims at minimiz-ing the energy consumption of any given process.Often however, one is interested in a performancemeasure that can incorporate all contributions tothe cost of a water recovery system including weight,power, volume and cost of supply. Thus, the equiv-alent cost (1) is employed as a multiobjective per-formance criterion. This performance index allows

Temp"C

80

60

40

20

0

-

-

—

1 1 1Hot Utll. -

//Hot CUI-TC / 1

/ 1f 1 Cold Curre

Cold Util.

1 1 1

-0.5 0.0 0.5 1.0 1.5

-

-

_

2.

Heat Load 10s J/hr

Figure 6: Composite Curve Diagram corresponding toS8

comparison of different flowsheets, including hybridones.

The structure under investigation consists of atwo step purification procedure. The first step iscomprised of VCD and/or TIMES and/or Air Evap-oration (AE). The final purification step consists ofa Multifiltration Unit (MF). For VCD, TIMES andMF weight, volume and power requirements werecalculated based on the information given in [10]and [18]. For AE, power consumption was calcu-lated based on the mathematical model and thesubsequent subsystem optimization that was dis-cussed earlier. The weight (Wc) of the wick evap-orator and the heat exchangers were calculated bythe same optimization procedure. The weight ofthe remaining equipment (piping, heat pump's com-pressor, packaging etc.) was considered to dependon the waste stream flowrate. Three such relationswere employed:

1. We = 15 kg

2. Wc = 35 kg

3. Wc = (17x2-|-3x+15) kg

where x = waste flowrate/(33kg/day). For a nineman crew the value of x does not exceed 1. In thiscase, the constant weight estimate represents up-per and lower values of the weight of the AE unitestimated by the third formula.

The cost of the three competing technologies(VCD, TIMES and AE) is shown as a function ofthe duration of the mission for a nine man crewin Figure 7. The mission duration affects the totalamount of power consumption and the equivalent

cost of the volume of the equipment. By observa-tion it can be seen that depending on the lengthof the mission different technologies may be moreattractive. The low specific energy characteristicsof VCD make it more attractive for long missionswhile for shorter missions TIMES and AE demon-strate lower values of the equivalent cost index. InFigure 8, the effect of the weight estimate for theAE unit is quantified. In particular, reducing theweight to 15 kg the corresponding curve drops by aconstant amount. This change in the relative posi-tion of the TIMES and AE curves (Figure 8) sug-gests that use of AE is advantageous over use ofTIMES over a longer mission duration interval. Forthe case of a three man crew the same calculationswere performed and are summarized in Figure 9.Due to the lower feed flowrate to the correspondingprocesses, TIMES is superior to VCD over a widermission duration interval. For the same reason, theweight estimate chosen for AE has a greater effecton the equivalent cost of the unit than it did in thenine man crew case.

S2<0

"oQtoo

8o

1 2 3 4 5 6Million Tlma (Years)

Figure 7: Cost Dependence of Mission Length for 9-Man Crew

The calculations described in the previous para-graph provide significant insight in the process se-lection procedure when only one technology is em-ployed in the first step of the flowsheet under consid-eration. However, there is another important alter-native that need be explored: do schemes utilizingmore than one competing technologies in series orin parallel achieve better performance than singletechnology schemes? In this direction a hybrid sys-tem of TIMES and VCD in parallel followed by anMF unit was considered. The equivalent cost for theoverall system as a function of the feed fraction pro-cessed through VCD for a nine man crew is shown

6 n—e—Air Evap (15)-a— Air Evap (35)- -O- -Times

2 3 4Mission Duration (Years)

Figure 8: Impact of Fixed Weight on System Cost forAE(9-Man Crew)

t.s

ID

O

a5

O.S

Air Evap (35)imes

-•o--Air Evap (15)•--•••- VCD

0 1 2 3 4 5 6Mission Duration (Years)

Figure 9: Variation of Cost with Mission Length for a3-Man Crew

for different missions in Figure 10. One should no-tice jump discontinuities near feed fraction valuesequal to 0.0 and 1.0. This occurs since the exis-tence of one unit is associated with a nonzero weightand volume. For small mission durations this factprohibits the use of any hybrid system. As the du-ration of the mission increases, the equivalent costcurves show a minimum which for missions longerthan 48 months suggests that a hybrid system si-multaneously utilizing VCD and TIMES in paral-lel is advantageous from an equivalent cost point ofview over a single technology system.

Hybrid system calculations were also performedin the case of TIMES and AE. Figure 11 depictsthe corresponding equivalent cost curves. The sen-sitivity of the equivalent cost value to the selectedweight estimate for the AE unit is also shown in thesame Figure. In particular, when one employs thequadratic correlation (Wc = 17i2 + 3i-M5) the cal-

u.su

0.45

0.40

Cost °-35

(I° * 0.30

0.25

0.20

0.

3mos' 6mos

9mw12mo318mn

•

)0 0.20 0.40 0.60 0.80

-

j

l.CFraction of Feed to VCD Processor

1.20

1.10

1.00

0.90

Cost 0.80(1°T> o.ro

0.60

0.50

0.40

0.30

24mos •48mo9 •72mos •96mos •

' liOmos •

0.00 0.20 0.40 0.60 0.80Fraction of Feed to VCD Processor

1.00

Figure 10: Equivalent Cost for VCD/TIMES HybridSystems

1.3

1

1 ''2

(O

O 1.1

I •0.9

0.8

• Times+ (15)o Times 4- (35)• Times + (quad)

1 : • . .o

•

ZO 40 60% Mr Evaporator

80 100

Figure 11: Hybrid TIMES/Air Evaporator Systems:Effect of AE Piping and Compressor Weight Correlationson the Cost (9-man Crew 6 months)

culations predict the existence of a minimum thatcorresponds to a hybrid configuration.

CONCLUSION

In this paper, we have developed mathemati-cal models for the design of optimal water recla-mation systems for long-duration space missions.These models were utilized in evaluating the per-formance of individual water reclamation technolo-gies. Two subsystems were also simultaneously op-timized with regard to the multiple objectives ofvolume, mass and power consumption. These ob-jectives were quantified through an operational costwhich was a linear function of these objectives. Theoptimization of the VCD and AE subsystems showsthat the appropriate choice of operating parametersleads to significant improvements of each technol-ogy.

It is then demonstrated that total mission costreductions can also be achieved by designing a waterreclamation system that consists of multiple tech-nologies. An optimization problem is formulatedfor this purpose based on the mathematical modelsdeveloped above. Two types of design techniqueswere employed: heat integration and multiobjectiveoptimization. For heat integration, pinch analysis isused to maximize heat recovery across three subsys-tems. The results show up to 94% reduction in heat-ing or cooling requirements over a non-integrateddesign. Multiobjective optimization shows that con-sidering a combination of technologies in the ECLSScan lead to cost reductions for long-range missions.

This work suggests that many options, beyondseries and parallel configurations exist, for an overalloptimal water reclamation system. Hence, a sys-tematic methodology, capable of searching for theoptimal flowsheet over all possible hybrid configu-rations of such systems, needs to be developed. Re-search efforts in that direction are currently under-way.

Acknowledgment

Financial support from NASA under grant no.NAG 2-822 is gratefully acknowledged.

Nomenclature

= weight of system (g)= power need for system (W)PT

VT = volume of system (m3)Tm = duration of UseCw = cost per unit mass ($/g) = $10/gCp = cost of power ($/KW-hr)

= $210/KW-hrCv = cost per unit volume per time

($/m3-t) = 31,800/m3-monthF = molar flow rate of airXwo = mole fraction of water in air phase

at wick outletXw = mole fraction of water in air phase

at wick inetXw0f=

m°le fraction of saturated waterin air phase at wick inet

Tun = temperature of water in air phaseat wick inlet

Twia = saturation temperature of waterin air phase at wick inlet

Two = temperature of water in air phaseat wick outlet

Twot = saturation temperature of waterin air phase at wick outlet

Td = temperature of water in air phaseat condenser inlet

L = length of wick evaporatorTfi = temperature of stream entering fanXho — mole fraction of water in air phase

at heater outletQc = heat removal in condenserQg = heat removal in heaterfobj = objective function

References

[1] NASA Research Announcement "Advanced lifesupport technology," NRA 2-34944 (LML),1992

[2] Dall-Bauman, L. and N. K. Rohatgi., "Systemsanalysis for sirf test bed," NASA 1992 LifeSupport Systems Analysis Workshop, Houston,Texas, 1992.

[3] Dehner, G. F., H. E. Winkler, and R. P. Reysa,"Thermoelectric integrated membrane evapo-ration subsystem operational improvements,"The Fourteenth Intersociety Conference on En-vironmental Systems, SAE Paper No. 840934,San Diego, CA, 1984.

[4] Ferrall, J. F., N. K. Rohatgi, and P. K. Seshan,NASA/JPL life support systems analysis sim-

ulation tool, Jet Propulsion Laboratory, Hous-ton, Texas, June 1992.

[5] Flynn, M., J. Fisher, and M. Hightower, Eval-uation of a potential regenerative water recla-mation subsystem design using the ASPENPLUS simulation program, NASA AMES RE-SEARCH CENTER(Advanced Life SupportDivision), 1991.

[6] Herrmann, C. C. and T. Wydeven, "Phys-ical/chemical closed-loop water-recycling forlong-duration missions," SAE Paper No.901446.

[7] Humphrey, J. L. and A. F Seibert, "Separa-tion technologies: An opportunity for energysavings," Chem. Eng. Prog., 88(32), 1992.

[8] Roebelen, G. J. Jr., G. F. Dehner, and H.E. Winkler, "Thermoelectric integrated mem-brane evaporation water recovery technology,"The Twelfth Intersociety Conference on Envi-ronmental Systems, SAE Paper No. 82849, SanDiego, CA, 1982.

[9] Manousiouthakis, V. "Multiobjective opti-mization of hybrid regenerative life supporttechnologies(annual report)," University ofChemical Engineering, Los Angeles, 1994.

[10] Miernik, J., B. Shah, and C. McGriff,"Wastewater processing technology for spacestation freedom: Comparative test data analy-sis," SAE Paper No. 911416, 1991.

[11] Morasko, G., D. Putnam, and R. Bagdigian,"Air evaporation closed cycle water recoverytechnology - advanced energy savings designs,"SAE Paper No. 860987, 1987.

[12] Otsubo, K., T. Tanemura, K. Nitta, M.Oguchi, H. Kashiwagi, and Y. Sawada, "Wastewater purification method using vapor com-pression distillation," Proc. 7-th InternationalSymposium on Space Technology and Science,11(1547-1554), Tokyo, 1990.

[13] Putnam, D. F. and R. Madsen, "An efficient airevaporation urine processing system for spacestation," SAE paper No. 881034, 1988.

[14] Putnam, D. F. and E. Thomas, "Recovery ofpotable of water from human urine," AerospaceMedicine, 736-739, 1969.

[15] Putnam, D. F., C. D. Jolly, G. V. Colombo,and D. Price, "Recovery of hygiene water bymultifiltration," The Ninteenth IntersocietyConference on Environmental Systems, SAEPaper No. 891445, San Diego, CA, 1989.

[16] Putnam, D. F., W. F. Michalek, and T. V.Pelt, "Space station hygeine water reclama-tion by multifiltration," The twenty first Inter-society Conference on Environmental Systems,SAE Paper No. 911553, San Diego, CA, 1991.

[17] Ray, R., S. Retzlaff, L. Radke-Mitchtll, D.Newbold, and D. Price, "A membrane-basedsubsystem for very high recoveries of spacecraftwaste waters," SAE paper No. 860984, 1987.

[18] Rohtagi, A., M. Ballin, P. Seshan, J. Ferrall,and V. Bilardo, " Hardware scaleup proceduresfor p/c life support systems," SAE Paper No.911396,1991.

[19] Schmidt, R. N., "Water recovery by vapor com-pression distillation," SAE Transactions, 433-437, 1989.

[20] Trusch, R. B. and Jr. G. J. Roebelen, "Athermoelectric integrated membrane evapora-tion system," The Intersociety Conference onEnvironmental Systems, ASME Paper No. 78-ENAs-19, San Diego, CA, 1978.

[21] Winkler, H. E. and Jr. G. J. Roebelen, "De-velopment of a preprototype thermoelectric in-tegrated membrane evaporation subsystem forwater recovery," The Intersociety Conferenceon Environmental Systems, ASME Paper No.80-ENAs-46, San Diego, CA, 1980.