Reliability Analysis of Experiment and Simulation Data for an Integrated Water Recovery System

Christian DouglassGeneral EngineeringUniversity of Illinois

Overview

• Problem: Can we test the reliability of life support systems before launch? Why has it been so difficult to test reliability in the past?

• Possible Solution: Crop reliability models developed, but how robust?

• Testing the solution: Crop reliability models are applied to wastewater experiment data and simulation data.

Problem • Physical means of early reliability testing

• High costs associated with testing

• Systems need to be tested until failure

• Mathematical and simulation models for early reliability testing

• Lower costs

• Systems can be tested until failure over and over

Possible Solution: Crop ReliabilityCan we model crop reliability after economic supply and demand?

S

D

Reliability Indicator,

0 DSR

)...,( 1 kXXtfS

)...,( 1 kXXtgD

Crop Reliability

k

iiiXaaY

10

k

i

biiXbY

10

Potato crop-system model in terms of response variable Y and predictor variables Xi :

Crop Reliability

Response Variable Y

Potato Leaf Dry Weight (after 90 days)

Predictor Variables

X2

X1

X3

X4

X5

CO2 concentration

Photoperiod

Photosynthetic photon flux

Temperature

Relative humidity

Crop Reliability

Possible Solution: Crop ReliabilityCan we model crop reliability after economic supply and demand?

S

D

Reliability Indicator,

0 DSR

)...,( 1 kXXtfS

)...,( 1 kXXtgD

Possible Solution: Crop ReliabilityCan we model crop reliability after economic supply and demand?

S

D

Reliability Indicator,

0 DSR

)...,( 1 kXXtfS

)...,( 1 kXXtgD

Taken from Kortenkamp, D. and Bell, S., “Simulating Advanced Life Support Systems for Integrated Controls Research,” Proceedings International Conference on Environmental Systems, SAE paper 2003-01-2546, 2003.

Testing the Model: the iWRS

• The iWRS is composed of four major subsystems:

• Biological Water Processor (BWP)• Reverse Osmosis (RO) System• Air Evaporation Subsystem (AES)• Post Processing System (PPS)

Testing the Model: the iWRS

Testing the Model: the iWRS

Goal: For each subsystem,

• Response variables

• Predictor variables

YQuantity

YQuality

Xi

• Potential Quantity Response Variables (PPS)

• Flow-meter (fm10)

•Tank weight scale (wt07)

• Potential Quality Response Variables (PPS)

• Total organic carbon sensor (toc)

• Dissolved oxygen sensor (do02)

Testing the Model: the iWRS

• Potential Predictor Variables (PPS)

• Temperature sensors

• Conductivity sensors

• Pressure transducers

• Valve states

Testing the Model: the iWRS

Different sampling

times

Binary sensor values

iWRS Problems

BioSim Life Support Simulation Modeling Tool

• Developed by NASA

• XML configuration files

• Java controllers

Testing the Model: BioSim

Testing the Model: BioSim

BioSim Problems

• VCCR module air exchange fixed

• OGS stochastic performance:

WaterRS Potable H2O Outflow Rate OGS Potable H2O Inflow Rate

Predictor Probability Distributions

Future Work

• Continue to explore possibility of using the iWRS experiment data

• Fix stochastic performance of OGS module

• Continue to find probability distributions for BioSim predictor variables

• Begin regression analyses of BioSim log data

Acknowledgements• Advisors Haibei Jiang and Professor Luis Rodríguez• Undergraduate research assistants Izaak Neveln and David Kane• Graduate student Glen Menezes• BioSim developer Scott Bell• The Illinois Space Grant Consortium• NASA grant No. NNJ06HA03G• The Boeing Company• The Aerospace Engineering Department• The Agricultural and Biological Engineering Department