Integrated modeling activities for the James Webb Space Telescope:

Structural-Thermal-Optical Analysis
John D. Johnstona, Joseph M. Howarda, Gary E. Mosier,a Keith A. Parrisha
Mark A. McGinnis, b A. Marcel Bluth,b Kevin Kim, b and Kong Q. Hac
a
NASA Goddard Space Flight Center, Greenbelt, MD
b
Swales Aerospace, Beltsville, MD
c
Jackson and Tull, Chartered Engineering, 7375 Executive Place, Seabrook MD

ABSTRACT

The James Web Space Telescope (JWST) is a large, infrared-optimized space telescope scheduled for launch in 2011.
This is a continuation of a series of papers on modeling activities for JWST. The structural-thermal-optical, often
referred to as "STOP", analysis process is used to predict the effect of thermal distortion on optical performance. The
benchmark STOP analysis for JWST assesses the effect of an observatory slew on wavefront error. Temperatures
predicted using geometric and thermal math models are mapped to a structural finite element model in order to predict
thermally induced deformations. Motions and deformations at optical surfaces are then input to optical models, and
optical performance is predicted using either an optical ray trace or a linear optical analysis tool. In addition to baseline
performance predictions, a process for performing sensitivity studies to assess modeling uncertainties is described.

Keywords: JWST, thermal distortion, optics, integrated modeling, image quality, telescope

1. INTRODUCTION
The James Webb Space Telescope is a large, near- and mid-infrared optimized space telescope under development by a
team consisting of NASA, a prime contractor team led by Northrop Grumman Space Technology (NGST), the
European Space Agency (ESA), and the Canadian Space Agency (CSA). Development of the JWST is led by the JWST
project at NASA Goddard Space Flight Center (GSFC). JWST will have an 18-segment, 6.5-meter primary mirror and
will reside in an L2 Lissajous orbit. The observatory, Fig. 1, is composed of three main elements: an Optical Telescope
Element (OTE), an Integrated Science Instrument Module (ISIM), and a Spacecraft consisting of the spacecraft bus and
sunshield. The OTE is a deployable cryogenic optical system with diffraction limited performance at 2 µm. The OTE
consists of the primary mirror segment assemblies (PMSAs), a secondary mirror support structure (SMSS), an aft optics
assembly (AOS), and a backplane structure that supports the preceeding subsystems. The ISIM consists of three science
instruments (Near-infrared camera, Near-infrared multi-object spectrograph, Mid-infrared camera) and a fine guidance
sensor all of which are mounted to a common ISIM bench structure that is kinematically mounted to the OTE
backplane. The spacecraft provides pointing and housekeeping functions for the observatory, while the deployable
sunshield provides passive radiative cooling and stray light control for the telescope and science instruments.

Figure 1: JWST Observatory Architecture

600 Optical, Infrared, and Millimeter Space Telescopes, edited by John C. Mather,
Proceedings of SPIE Vol. 5487 (SPIE, Bellingham, WA, 2004)
0277-786X/04/$15 · doi: 10.1117/12.551704

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 02/25/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 2. INTEGRATED MODELING

Multi-disciplinary engineering analysis, or integrated modeling, is a critical element of the JWST mission. Integrated
modeling primarily supports engineering design and verification of high level optical requirements for image quality
and sensitivity. The image quality requirements are specified by Strehl ratio and encircled energy. Current plans for
end-to-end optical performance verification rely on modeling to a degree surpassing previous programs. The final
verification process requires modeling and analysis to correct for the effects of the optical test procedure, metrology,
gravity, and thermal/seismic effects on opto-mechanical stability before extrapolation to behavior in the on-orbit
environment.

Performance (error) budgets are maintained to control allocations to the various subsystems that comprise the JWST
observatory. These budgets are traceable to the high level optical requirements. The image quality requirements are
recast as requirements on wavefront error. These allocations may be roughly classified as belonging either to calibration
(alignment and figuring) of the optics or to opto-mechanical stability between periodic recalibrations. The sensitivity
requirement is the basis for allocations to radiometric performance, stray light suppression, and detector performance.
One could view the primary integrated modeling activities as consisting of four distinct multi-disciplinary analysis
efforts, three supporting verification of image quality requirements, and the fourth supporting verification of the
sensitivity requirement. The analyses supporting verification of the image quality requirements are (1) thermal
distortion, or STOP (Structural-Thermal-Optical), to estimate alignment figure drift due to observatory re-pointing and
other transient factors, (2) jitter, to estimate the blurring and distortion due to uncompensated pointing and vibration,
and (3) wavefront sensing and control, to estimate the post-calibration alignment and figure errors.

The applications of integrated modeling will change over the program life cycle. In the formulation and requirements
definition phase, a strawman design was developed to address the high-level mission requirements, goals, and
constraints. The role of integrated modeling was to validate this design concept by showing that the conceptual design
met the requirements with margin, subject to reasonable assumptions, and that initial sub-allocations to observatory
elements and sub-systems were also reasonable. Following a series of requirements reviews at the various program
levels (mission, observatory, telescope, instruments), the modeling activity has aligned with the architecture/design
activity in a series of cycles, of 6-9 month duration. There will be multiple design cycles between major program review
milestones. At the beginning of each cycle, a baseline design (or several as long as significant design trades are active)
will be “frozen”, and the set of multi-disciplinary analyses will be executed to verify that the baseline design(s) for that
cycle meet the optical system requirements, with margin. The analysis will not only produce predictions of nominal
design performance, but will also address uncertainties in performance due to variability in design parameters, material
properties, and the environment.

The focus of this paper is on structural-thermal-optical integrated performance analysis for JWST. The STOP analysis
process will be described, followed by representative results based on models representing the design at the time of the
system requirements review (SRR). The paper concludes with a discussion of planned STOP analysis sensitivity
studies.

3. STRUCTURAL-THERMAL-OPTICAL (STOP) ANALYSIS PROCESS

3.1. Overview
The performance budget dictates that the opto-mechanical stability of the JWST observatory be maintained within
specification between recalibrations. The current performance budget allocates 25 nm,rms WFE for alignment drift and
32 nm,rms for figure drift. This leads to a RSS total allocation of 41 nm,rms for drift stability. One of the uses of STOP
analysis on the JWST project, and the focus of this paper, is to estimate alignment figure drift due to thermal distortions
resulting from re-pointing (slewing) of the observatory. The multi-disciplinary STOP analysis process described here
links thermal, structural, and optical models in a “bucket-brigade” fashion to predict the response of the system to a
slew maneuver about the pitch axis. The process is outlined schematically in Figure 2. The focus of current analyses is
to predict the change in optical performance due to a “worst-case” slew between the hot and cold observatory
orientations. This slew case results in the maximum thermal distortion and hence bounds the expected performance
range. The process begins with a thermal analysis to determine steady-state temperature distributions for a range of

Proc. of SPIE Vol. 5487 601

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 02/25/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 observatory orientations. The orientations corresponding to hot and cold cases are identified, and temperature results are
“mapped” from the thermal to the structural model. The structural model is then used to predict distortions resulting
from thermal loading, and displacements of optically important surfaces are “mapped” to the optical model. Finally,
perturbations from the structural analysis results are introduced to the optical model and the optical performance of the
system is predicted. Further details regarding the discipline models and analyses are provided in the following sections.
The models described here were developed by the prime contractor team and delivered to the government team at the
time of the observatory system requirements review.

Observatory Thermal Temperature

Thermal Transient vs Time
Models Analysis (thermal)
TSS and SINDA,
Map
or IDEAS/TMG Temperatures
Temperature to Structure
PATRAN
vs Time EXCEL
(structural)

Observatory Displacements
Structural Static Loads of optical Prelim. Linear Optical
Model Analysis elements and Tool
NASTRAN surfaces MATLAB

Final

Wavefront
Generate Optics Model Ray-Trace Error
Interferograms Analysis (OPD Map &
SigFit OSLO RMS value)

Figure 2: STOP analysis process map

3.2. Thermal model

The JWST observatory thermal model consists of a TSS radiation model and a SINDA/G thermal network math model.
The TSS model consists of 3 major parts. The first part is the geometry file. This file defines shape, size, location, and
surface properties. The second part is the radk file. By using the geometry file and ray tracing methods, radiative
conductors are calculated and generated. The last part of TSS is a heat rate run. This file calculates and generates heat
rates related to solar and infrared energy. Thirteen different cases were considered based on the sun angles
corresponding to different observatory orientations. Eleven cases are based on sun angle (+5° to –45°) based on rotating
about pitch axis. The two cases are based on sun angle (+/-5°) around the roll axis. SINDA/G is used to predict the
temperature of the system using data set from the TSS runs, user generated linear conductive coupling, and necessary
heater/power dissipations. Only the steady state response is considered here.

The geometry of the thermal model consists of several components. The first component is cold external geometry of
the JWST observatory as shown in Figure 3. The second part is internal geometry that includes details for the primary
mirror (PM) and the backplane (BP) assembly. The third major component is the 5 layers of sunshield. The fourth
component is hot side of JWST that includes sunshield layer 1, solar array, bus and closeouts. The final piece of the
component is internal details of deployable tower. Note that the optical properties used in all thermal models are end of
life (EOL) properties.

Based on temperatures obtained from the thermal math model and the model topology contained in the geometric
thermal model, a table of coordinates and corresponding temperatures is obtained for hand-off to the structures
discipline. There are several approaches available for mapping the thermal model results to the structural model. One
approach is to fit a function to the thermal results, then map this function to the structural model mesh. Another

602 Proc. of SPIE Vol. 5487

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 02/25/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 approach pairs corresponding thermal and structural model nodes (within a specified geometric tolerance) and assigns
these nodes the same temperatures. These temperatures are then prescribed as “seeds” in a steady-state heat transfer
analysis performed using the finite element model that provides a complete set of nodal temperatures for the structural
analysis. Thermal mapping was accomplished via the first process for the analysis discussed in this paper.

Figure 3: Thermal model

3.3. Structural model

The NASTRAN structural model of the deployed observatory is shown in Figure 4. The fidelity of the model, which has
approximately 40000 degrees of freedom (DOFs), is appropriate for this stage of the project. It contains sufficient detail
in the optical support structure areas to represent important thermo-structural effects without undue detail that would
slow trade studies. One-dimensional beam elements are used to model the backplane, backplane support frame, and
ISIM bench. Note that in actuality the OTE and ISIM metering structures are entirely formed of composite orthotropic
material systems. Subsequent models will account for the thermal-structural response characteristics in all principle
material directions. The primary mirror segment models are sufficiently detailed to capture critical resonant modes but
are not appropriate for detailed stress analysis. Currently a single temperature is calculated for each segment in the
thermal models, therefore more detailed structural models would not be useful. The mathematical validity of the STOP
model is assured by means of a series of standard model check analyses. Unit gravity loading in each axis, unit-enforced
displacement (and rotation) in each axis, ungrounded normal modes, and unit temperature change checks were run on
the model.

Temperature-dependent values for the coefficient of thermal expansion are included in the model. NASTRAN requires
that the tabulated values are “secant” CTE values. Element thermal strain is calculated by multiplying a single CTE
table entry by the difference between the applied temperature and the material reference temperature. The thermal strain
behavior of all materials is modeled in this way. This implementation of temperature-dependent thermal strain in
NASTRAN complicates analyses between two on-orbit states, such as a slew to a new observation target. Two
successive linear static analysis runs, both from the reference temperature state, are required. The applied temperature
loads are the initial and final on-orbit temperature states. The difference between these two states represents the effect of
the slew. The differences between states are calculated outside of NASTRAN, such as in pre- and post-processing
software or MATLAB.

Structural deformations are input as perturbations to the optical model to assess optical performance. Mappings were
required between the structural model and two different optical models (see discussion in the following section): optical
ray trace model and linear optical model. An interferogram file is used as the interface between the NASTRAN
structural model and the OSLO optical ray trace model for mapping primary mirror deformations. The interferogram
file is generated using SigFit software developed by Sigmadyne, Inc. (Version 2003-R2). The interferogram (or hitmap)
is a rectangular array of data over an optical surface onto which surface deformations are interpolated. Note that for this
analysis a single interferogram was generated for the entire primary mirror and applied as a perturbation to an optical
model with a monolithic primary mirror. Alternately, interferograms could be generated for each segment of the

Proc. of SPIE Vol. 5487 603

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 02/25/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 primary mirror assembly and used as perturbations in a segmented optical model. Additionally, since the current
structural model only provides predictions for optical surface deformation on the primary mirror, rigid body motions are
provided to the optical analyst for the secondary mirror, tertiary mirror, fine steering mirror, and image plane. The
interface between the structural model and the linear optical model is a matrix of six degree of freedom motions, three
translations and three rotations, for 22 optics (18 primary mirror segments, secondary mirror, tertiary mirror, fine
steering mirror, and image surface).

Telescope

Spacecraft
ISIM (hidden) Tower/Isolator /Sunshield

Figure 4: Structural model

3.4. Optical models

The JWST government analysis team uses two different models to assess optical performance. The first is an optical ray
trace model, and the second is a linear optical model that is used primarily for first-order perturbation studies. The
following sections describe each of the models in detail.

3.4.1 Optical ray trace model

Ray tracing for this analysis is performed using either of two optical prescriptions delivered from the prime contractor:
one having a monolithic primary mirror surface, and the other having a segmented primary (non-sequential surface).
The former is generally used for analyses that assume the primary mirror is aligned perfectly, while the latter is used to
more accurately model alignment procedures and other more detailed analyses. Both models are delivered in OSLO
optical design software lens format. These delivered models are modified slightly to decouple the coordinate systems
between optical surfaces to convert the sequential layout coordinates to a global coordinate system. This allows each
surface to be moved independently with respect to all other surfaces, so that any secondary mirror motion, for example,
does not also move the tertiary mirror and other follow-on surfaces.

Rigid body perturbations due to thermal (or other) loads are applied to either optical model by changing the coordinates
defining the location of each optic, while deformations to the optical surfaces are applied to each surface using an
interferogram file, a grid of normal displacements from the optical surface. When the monolithic primary mirror model
is used, a combined interferogram of all primary mirror segments are contained within a single file, which then
simulates a segmented model to the level of the grid spacing of the interferogram.

The primary figure of merit for this analysis is root-mean-square wave-front error (RMS WFE), evaluated at the exit
pupil of the system for the central field point. This is also referred to as a map of optical path differences, or “OPD
map”. If the rate of change of the perturbation is much slower than the exposure time or the ability for the fine guidance
system to track the image, as it will be for typical thermal influences, then the RMS WFE is evaluated with the best-fit
plane removed. This figure of merit is calculated in OSLO, but we also export the opd map into MATLAB for further
analysis and archiving.

604 Proc. of SPIE Vol. 5487

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 02/25/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 3.4.2 Linear optical model
The approach taken for the linear optical model involves performing all ray tracing in advance through the system with
a single known rigid body perturbation applied to each surface and each degree of freedom. This generates linear
sensitivities, which are used as transfer matrices converting rigid body motions of optical components to absolute image
motion at the detector and wave front error induced due to the misaligned components. The linear optical model can be
briefly described as a first-order Taylor expansion of the optical path lengths of a grid of rays traced through the system,
uniformly spaced at the entrance pupil. The variables expanded upon are the rigid body degrees of freedom (DOF) for
each optical component in the system. Naturally, the model is valid in the linear approximation only for small
perturbations of the rigid body degrees of freedom. This can be summarized in the following equation:
L = L0 + ∂∂UL ∆U + O ( 2 ) (1)
where L contains the optical path lengths of the grid of rays from the entrance pupil to the exit pupil, L0 contains the
nominal (unperturbed) path lengths for the system, ∆U represents the perturbation of the rigid body degrees of
freedom, and O (2 ) represents the second and higher order terms of the expansion which are disregarded for this linear
∂L
analysis. The first-order derivative term, ∂U , is constructed by perturbing each DOF, one at a time, and evaluating the
change in path lengths from the nominal for each case. This process is described in detail in Ref. 1. Note that this first-
order derivative term contains all wavefront data including absolute piston, tip/tilt, and all higher-order terms, since it is
evaluated with respect to the nominal reference sphere at the exit pupil without regard to chief ray motion. Wavefront
error is defined here as the optical path length difference, or OPD, between the perturbed system and the nominal
system where all path lengths are equal to that of the chief ray. Comparing wave front maps generated from Eq. (1) to
those generated by ray tracing show less than 1% difference for random realizations of perturbations of all components
up to 1 micron or micro-radian motions or tilts, respectively.

Structural analysts evaluate L by determining elements of U resulting from deformations calculated in NASTRAN,
∂L
scaling the ∂U matrices by these vector components; and summing the products. The root-mean-square (RMS) value of
L is the key performance metric, although others are used. A number of MATLAB functions and scripts have been
created to facilitate the wavefront error calculation process. Among the capabilities added by these scripts are:
minimization of RMS OPD with the fast-steering mirror; slew maneuver modeling from the results of two NASTRAN
runs, including the ability to remove reference states; automated NASTRAN results plotting with performance metric
labeling; and decomposition of OPD results into spatial frequency components.

Primary Mirror
Secondary Mirror
Tertiary Mirror

Fast Steering Mirror Image Surface

Figure 5: Optical model

Proc. of SPIE Vol. 5487 605

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 02/25/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 4. STOP ANALYSIS RESULTS FOR JWST AT SRR

4.1. Thermal model results

Figure 6 presents the steady-state temperature profile for the hot case (-15° pitch angle) illustrating primary mirror (PM)
and backplane (BP) structure temperatures. The hot case is defined by the hottest average PM temperature among the
cases studied. The plot indicates that for the hot case the maximum PM temperature is 58.3 K and the minimum
temperature is 35.2 K. Note that the PM temperature gradient is primarily in the V3 direction with the hot side nearest
the sunshield. Also note that the backplane structure has a temperature gradient that is similar albeit at a slightly higher
temperature than the PM. Of interest in assessing the operational WFE stability of the telescope is the worst case
temperature change as the observatory slews about the pitch axis. Figure 7 presents a plot of the worst case delta
temperature profile (hot case – cold case), where the cold case is for a -45° pitch angle. The average change in
temperature between states is approximately 0.3 K. The maximum change in temperature between states is less than 0.5
K over the entire PM and BP with the peak change occurring nearest the sunshield.

Figure 6: Temperature profile (‘Hot’ case) plot

Figure 7: Delta temperature profile (‘Hot-Cold’ case) plot

606 Proc. of SPIE Vol. 5487

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 02/25/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 4.2. Structural model results
Figure 8 presents results from the structural model illustrating the deformations resulting from a change in temperature
corresponding to the worst-case slew defined by the thermal analysis. Peak total displacements are on the order of 200
nm and occur in the secondary mirror support structure. The magnitude of the total displacements exhibited by the
primary mirror (primary mirror segments plus backplane support structure) is on the order of 50 nm. Of interest in
assessing wavefront error are the primary mirror surface normal displacements. NASTRAN results were post-processed
using the Sigfit software package to obtain surface normal displacements with rigid body motions removed. The RMS
value of the surface normal displacements is 8 nm and the peak-to-valley value is 42 nm.

The structural model predicts that while the backplane support structure is the main contributor to primary mirror
deformations, the PMSA’s also make a significant contribution. Analysis shows that the deformations of the individual
PMSA’s are primarily due to thermal strains in the primary mirror support and actuation system with lesser
contributions from the mirror substrate. The deformations result in a rigid-body-like pistoning of each of the segment
mirror substrates with negligible surface deformation. Note that since the change in primary mirror temperature between
states is greater near the sunshield, the PMSA’s on that side of the primary mirror undergo larger motions than the
PMSA’s opposite the sunshield.

Figure 8: Structural deformation plot

4.3. Optical model results

Results from the optical models with perturbations based on structural deformations resulting from the worst-case slew
case are presented in this section. Optical performance predictions for RMS WFE are broken down into three spatial
frequency bands: low (< 5 cycles per aperture), mid (>5 and <35 five cycles per aperture), and high (>35 cycles per
aperture). These contributions are calculated in such a way that when RSS’s they equal the total WFE. Optical results
below are presented as total RMS WFE followed by the frequency decompositions.

Figures 9 and 10 present maps of the optical path difference evaluated at the exit pupil of the system for the central field
point. The OPD map from the optical ray trace model, Fig. 9, shows an RMS WFE of 18.2 nm (Low=17.8 nm, Mid=6
nm, High=2 nm) and a peak-to-valley WFE of 89 nm. The OPD map from the linear optical model, Fig. 10, shows an
RMS WFE of 14 nm (Low=13.2 nm, Mid=4.6 nm, High=1.6 nm) and a peak-to-valley WFE of 72 nm. Overall, results
from the two optical models show good agreement. Qualitatively the OPD maps are very similar, and the RMS WFE
predicted by the linear optical model is 4 nm less than that predicted by the ray trace. As noted in the structural results
section, the two main contributions to WFE are deformations in backplane support structure and the PMSA’s. On a

Proc. of SPIE Vol. 5487 607

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 02/25/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 global scale, the OPD is dominated by a power aberration associated with deformations in backplane support structure.
Examination of the WFE by spatial frequency band shows that, as expected, the WFE is dominated by a low spatial
frequency contribution. The analytical results from the ray trace and linear optical models show that the observatory
currently meets an RSS total allocation of 41 nm,rms for drift stability with ample margin. Future studies, outlined in
the following section, will explore the sensitivity of these nominal results.

37.1 nm
Frequency Content:
<5 c.p.a. - 17.8 nm
5-35 c.p.a. - 6.0 nm
>35 c.p.a. - 2.0 nm

P-V
88.8 nm

-51.7 nm
18.2 nm, RMS WFE
Figure 9: OPD map plot from OSLO ray trace

30.7nm
Frequency Content:
<5 c.p.a. - 13.2 nm
5-35 c.p.a. - 4.6 nm
>35 c.p.a. - 1.6 nm

P-V
71.7 nm

-41.1 nm
14.0 nm, RMS WFE

Figure 10: OPD map plot from linear optical model

608 Proc. of SPIE Vol. 5487

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 02/25/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 5. SENSITIVITY STUDIES

The sensitivity of observatory performance to variability in key parameters such as material properties and the
environment will be studied using Monte Carlo statistical models using changes in wavefront error as the scalar
parameter to correlate. The models will be applied at various levels of refinement to gain insight to the overall system
behavior and to define key parameters to consider in the analysis. The analysis flow will follow a process along the lines
of the following:
1. Define benchmark temperature cases to represent maximum expected disturbance inputs. Currently the benchmark
disturbance case is the worst-case slew between hot and cold pointing conditions.
2. Identify key variable parameters and a statistical model for each parameter. Expected parameters to consider
include coefficient of thermal expansion, thickness variations, and stiffness variations.
3. Determine the sensitivity matrix of WFE to each of the parameters by developing math models (engineering
formulas, finite element models, optical models, etc.) and studying the influence of each parameter on the
monitored response variable. Initial development of the sensitivity matrix will assume the parameters are
uncorrelated and can be linearly combined.
4. Generate random sets of each key parameter.
5. Combine each random set with the sensitivity matrix and calculate a random sampling of WFE due to variations in
key parameters.
6. Calculate the statistics (mean and variance) of the WFE variation and compare to the system requirements.

Variations of this approach will be used to study several items of interest including but not limited to the following:

Determine at the level of spatial variation for each parameter that needs to be considered. For example, does
the representation of the composite tubular elements in the OTE and ISIM structures require consideration of
all three material orthotropic axes? Is it sufficient to represent each tubular element with a uniform local
average or does the spatial variation need to be modeled at each element? These studies will support defining
the level of model fidelity required.

Study the effects of shifts or bias in the expected versus delivered nominal set point for each parameter
assuming the same variance model.

Study the impacts of localized bias in the assembled telescope elements. For example, the backplane center
section and wings could be fabricated from three different lots of composite prepreg. Also, the length to which
each actuator is extended will affect the athermalization for each bipod actuator set on each PMSA. These
conditions open the door for localized variations or bias in the nominal response of each parameter model.

Support manufacturing and process definition by evaluating various statistical models for each parameter,
defining the allowable variation, and identifying key parameters on which to monitor and control the process.

Study the effect of uncertainties in the thermal models and support the establishment of the required levels of
accuracy in thermal performance predictions.

Assess the level of uncertainty in nominal STOP analysis performance predictions.

Statistical models for thermal distortion of the observatory will be a valuable tool employed over the life of the project.
Initial models of low fidelity have already supported system requirements definition. Detailed component level models
are anticipated in support of each major review cycle, and an appropriately detailed system level model is required to
accomplish analysis on which the verification plan is anchored. The validity of these models will be supported by
testing and correlation activities at the material, component, and system levels.

6. SUMMARY
An overview of the integrated modeling process used by the JWST government team to predict the on-orbit thermo-
elastic stability of the observatory has been described. The multi-disciplinary analysis process involves linking thermal,
structural, and optical models to predict end-to-end performance. Results were presented for models representative of
the observatory concept at the time of the system requirements review in late 2003. Predictions from this analysis for
nominal performance indicate that the observatory concept meets current requirements for on-orbit stability. An
approach for assessing the sensitivity of the nominal performance due to variations in key parameters such as material
properties and thermal loading was outlined. Future studies will implement this process to support the further evolution
of the observatory design.

Proc. of SPIE Vol. 5487 609

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 02/25/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 REFERENCES

1. Joseph M. Howard, "Optical modeling activities for the James Webb Space Telescope (JWST) project: I. The linear
optical model," Proc. SPIE Int. Soc. Opt. Eng. 5178, 82 (2004).
2. Howard, J.M., et al, “Optical Modeling Activities for the James Webb Space Telescope (JWST): 2. Determining
Image Motion and Wavefront Error Sensitivities for a Segmented Optical System Over An Extended Field,” SPIE
Astronomical Telescopes and Instrumentation Conference, June 2004, SPIE 5487-142.
3. Hyde, T.H., et al, “Integrated Modeling Activities for the James Webb Space Telescope (JWST): Optical Jitter
Dynamics Analysis,” SPIE Astronomical Telescopes and Instrumentation Conference, June 2004, SPIE 5487-122.
4. Nella, J., “James Webb Space Telescope (JWST) observatory architecture and performance,” SPIE Astronomical
Telescopes and Instrumentation Conference, June 2004, SPIE 5487-23.
5. Mosier, G., Parrish, K., Femiano, M., Redding, D., Kissil, A., Papalexandris, M., Craig, L., Page, T., and Shunk,
R., "Performance Analysis Using Integrated Modeling", NGST Monograph No. 6, August 20006.
6. Mosier, G., Femiano, M., Ha, K., Bely, P., Burg, R., Redding, D., Kissil, A., Rakoczy, J. and Craig, L., “An
Integrated Modeling Environment for Systems-level Performance Analysis of the Next Generation Space
Telescope,” SPIE Vol. 3356, Space Telescopes and Instruments V, 1998.

610 Proc. of SPIE Vol. 5487

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 02/25/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx