14.12.2012 Views

Permissible Width of Front Face Flaws GMT Primary ... - Carnegie

Permissible Width of Front Face Flaws GMT Primary ... - Carnegie

Permissible Width of Front Face Flaws GMT Primary ... - Carnegie

SHOW MORE
SHOW LESS

Transform your PDFs into Flipbooks and boost your revenue!

Leverage SEO-optimized Flipbooks, powerful backlinks, and multimedia content to professionally showcase your products and significantly increase your reach.

<strong>GMT</strong> Project<br />

<strong>Permissible</strong> <strong>Width</strong> <strong>of</strong> <strong>Front</strong> <strong>Face</strong> <strong>Flaws</strong><br />

Doc # :<br />

Date: 2009-Feb-13<br />

Status: Rev. F<br />

Page: 1 <strong>of</strong> 9<br />

<strong>Permissible</strong> <strong>Width</strong> <strong>of</strong> <strong>Front</strong> <strong>Face</strong> <strong>Flaws</strong><br />

<strong>GMT</strong> <strong>Primary</strong> Mirror Off-Axis Segment<br />

Prepared By:<br />

Name(s) and Signature(s)<br />

Brian Cuerden<br />

Approved By<br />

Date<br />

Name and Signature Title<br />

PI<br />

Technical Division Manager<br />

Responsible Engineer<br />

Date<br />

Optical Engineer<br />

NOTE: THIS DOCUMENT CONTAINS PROPRIETARY INFORMATION AND CANNOT BE<br />

DISCLOSED WITHOUT THE WRITTEN CONSENT OF STEWARD OBSERVATORY.


Revision History<br />

<strong>GMT</strong> Project<br />

<strong>GMT</strong>, <strong>Permissible</strong> <strong>Width</strong> <strong>of</strong> <strong>Front</strong> <strong>Face</strong> <strong>Flaws</strong><br />

Doc # :<br />

Date: 2009-Feb-13<br />

Status: Rev F<br />

Page: 2 <strong>of</strong> 9<br />

Issue Date Changes Responsible<br />

A 23-Jan-2009 Initial release Brian Cuerden<br />

B 26-Jan-2009 Standardize<br />

Factor<br />

on Blain’s Geometry Brian Cuerden<br />

C 27-Jan-2009 Corrected Y axis label <strong>of</strong> Fig. 4.2<br />

Corrected 3sigma a/w to 0.4 (section<br />

3.2, Table 4.1 and sections 5 and 6)<br />

Brian Cuerden<br />

D 28-Jan-2009 Incorporated numerous comments from<br />

Blain Olbert<br />

Brian Cuerden<br />

E 4-Feb-2009 Corrected a/2w to a/w in fig. 4.1 & 4.2 Brian Cuerden<br />

F 13-Feb-2009 Standardized phrasing Buddy Martin


Table Of Contents<br />

<strong>GMT</strong> Project<br />

<strong>GMT</strong>, <strong>Permissible</strong> <strong>Width</strong> <strong>of</strong> <strong>Front</strong> <strong>Face</strong> <strong>Flaws</strong><br />

Doc # :<br />

Date: 2009-Feb-13<br />

Status: Rev F<br />

Page: 3 <strong>of</strong> 9<br />

1. Applicable Documents and Drawings ........................................................................ 4<br />

2. Acronyms and Abbreviations ..................................................................................... 4<br />

3. Introduction.................................................................................................................4<br />

3.1. Description <strong>of</strong> <strong>Front</strong> <strong>Face</strong> <strong>Flaws</strong>......................................................................... 4<br />

3.2. Calculation <strong>of</strong> the <strong>Permissible</strong> Flaw <strong>Width</strong> ........................................................ 5<br />

4. Discussion...................................................................................................................5<br />

4.1. Calculation Procedure......................................................................................... 5<br />

4.2. Use <strong>of</strong> Flaw <strong>Width</strong> as the Controlling Parameter ............................................... 7<br />

5. Results......................................................................................................................... 9<br />

5.1. <strong>Permissible</strong> Initial Flaw at 308 psi for One year life .......................................... 9<br />

5.2. <strong>Permissible</strong> Initial Flaw at 231 psi for One year life .......................................... 9<br />

5.3. <strong>Permissible</strong> Initial Flaw at 308 psi for Six Months............................................. 9<br />

6. Conclusions................................................................................................................. 9


<strong>GMT</strong> Project<br />

<strong>GMT</strong>, <strong>Permissible</strong> <strong>Width</strong> <strong>of</strong> <strong>Front</strong> <strong>Face</strong> <strong>Flaws</strong><br />

1. Applicable Documents and Drawings<br />

Doc # :<br />

Date: 2009-Feb-13<br />

Status: Rev F<br />

Page: 4 <strong>of</strong> 9<br />

The following documents <strong>of</strong> the exact issue shown form a part <strong>of</strong> this Specification to the<br />

extent specified herein. In the event <strong>of</strong> conflict between the documents referenced herein<br />

and the present document, the terms <strong>of</strong> the Specification shall be considered as<br />

superseding requirements.<br />

Applicable Documents<br />

[AD1] <strong>Face</strong> Plate <strong>Flaws</strong>, Rev. D<br />

[AD2] Weiderhorn and Roberts, “Fracture Mechanics Study <strong>of</strong><br />

Skylab Windows, NBS Project 3130450, NBS Report 10 892,<br />

05/31/1972<br />

[AD3] <strong>GMT</strong> <strong>Face</strong>plate Life Final R00.pdf<br />

Drawings<br />

2. Acronyms and Abbreviations<br />

<strong>GMT</strong> Giant Magellan Telescope<br />

TBR To Be Reviewed<br />

a The depth <strong>of</strong> a flaw<br />

KI The flaw tip stress intensity<br />

KIC The critical flaw tip stress intensity<br />

w The full width <strong>of</strong> a flaw<br />

Y The flaw tip stress intensity geometry factor<br />

3. Introduction<br />

3.1. Description <strong>of</strong> <strong>Front</strong> <strong>Face</strong> <strong>Flaws</strong><br />

Visible front face flaws are located near the center <strong>of</strong> the mirror. Most are<br />

arranged in bands 2.0 inches apart, the remainders are in bands 1.0 inch apart.<br />

Generating was performed with a 2.0” spiral cut with finish cuts being 1.0”<br />

apart. Etching apparently flaw free areas along tracks <strong>of</strong> visible flaws or<br />

stepped 2.0” laterally from these tracks has revealed additional flaws.<br />

Additional areas are being/have been etched and a separate report will<br />

describe the results <strong>of</strong> this flaw detection effort which will determine the


<strong>GMT</strong> Project<br />

<strong>GMT</strong>, <strong>Permissible</strong> <strong>Width</strong> <strong>of</strong> <strong>Front</strong> <strong>Face</strong> <strong>Flaws</strong><br />

Doc # :<br />

Date: 2009-Feb-13<br />

Status: Rev F<br />

Page: 5 <strong>of</strong> 9<br />

width <strong>of</strong> the 99 th percentile flaw expected anywhere on the remaining, unetched,<br />

mirror face.<br />

3.2. Calculation <strong>of</strong> the <strong>Permissible</strong> Flaw <strong>Width</strong><br />

4. Discussion<br />

This report defines the permissible flaw width on the front surface. It uses twice<br />

the estimate <strong>of</strong> the maximum front face stress defined in reference [AD1]. This is<br />

2*154 = 308 psi. This stress is assumed to be applied for an accumulated total <strong>of</strong><br />

1 year in the 50 year operational life <strong>of</strong> the primary mirror. This time <strong>of</strong> exposure<br />

is intended to represent a worst case estimate <strong>of</strong> the exposure time. A naïve<br />

estimate <strong>of</strong> exposure time would be to assume that the worst case thermal stress<br />

occurs for 6 hours every month on those nights when the mirror is warmer than<br />

the night ambient (reported to occur after several days <strong>of</strong> downtime). This<br />

represents 0.83% <strong>of</strong> the time or 0.42 years in 50 years.<br />

Measurements <strong>of</strong> the geometry <strong>of</strong> revealed flaws give a mean depth, a, <strong>of</strong> 23% <strong>of</strong><br />

the width, w. Three times the standard deviation <strong>of</strong> the depth to width ratio, a/w,<br />

is 16%, making the 99.9 th percentile flaw 39% as deep as it is wide.<br />

4.1. Calculation Procedure<br />

In reference [AD1] the flaw tip stress intensity was computed using equations<br />

for long surface flaws, a/w


<strong>GMT</strong> Project<br />

<strong>GMT</strong>, <strong>Permissible</strong> <strong>Width</strong> <strong>of</strong> <strong>Front</strong> <strong>Face</strong> <strong>Flaws</strong><br />

Doc # :<br />

Date: 2009-Feb-13<br />

Status: Rev F<br />

Page: 6 <strong>of</strong> 9<br />

The differences are not significant since we are in good agreement on the<br />

initial dimensions <strong>of</strong> flaws which grow to failure in one year. In particular,<br />

the failure definition used by Olbert, [AD3], is when the flaw reaches critical<br />

size or grows through the faceplate. Cuerden allows the flaw to grow through<br />

the faceplate. In most cases this does not happen at 308 psi. When it does<br />

happen, it occurs very near the end <strong>of</strong> the one year period and has little effect<br />

on the result (the size <strong>of</strong> the initial flaw). One other difference is that Cuerden<br />

estimates the change in the aspect ratio <strong>of</strong> the flaw as it grows using the<br />

known flaw geometry parameters at the depth- and corner- points <strong>of</strong> the flaw.<br />

This has a small effect on the result, confirmed by freezing the aspect ratio<br />

and repeating the calculation. For an a/w=0.2 flaw with a growth-evolving<br />

aspect ratio, the permissible initial flaw size is 29.45 mm wide and 5.89 mm<br />

deep. When the aspect ratio is frozen at 0.2, the permissible initial flaw size is<br />

26.8 mm wide and 5.36 mm deep. Figure 4.1 shows the flaw depth and aspect<br />

ratio over time for the variable aspect ratio case. Figure 4.2 compares the flaw<br />

intensity geometry factor (the Y in KI = Y*σ√a)*. The reduction in the<br />

geometry factor as the flaw aspect ratio changes toward 0.4 over time gives a<br />

smaller geometry factor and therefore a reduced crack growth rate permitting<br />

a slightly larger initial flaw. Cuerden’s variable aspect ratio is based on<br />

evaluating crack growth at two points, one at the deepest point <strong>of</strong> the flaw<br />

and the other where the flaw meets the surface, assuming the flaw shape<br />

remains elliptical. It serves to demonstrate that the change in flaw geometry is<br />

beneficial but it may not be conservative. Olbert’s constant aspect ratio<br />

calculation is therefore used to establish permissible flaw widths.<br />

* Flaw propagation is a function <strong>of</strong> the flaw tip stress intensity, KI. This has<br />

the units force*sqrt(length)/unit area = stress*sqrt(length). The flaw tip stress<br />

intensity is generally proportional to the stress times the square root <strong>of</strong> the<br />

flaw length. Y is the proportionality constant which is a constant for some<br />

simple flaw configurations (for example a flaw in the edge <strong>of</strong> a semi-infinite<br />

plate) but varies with the flaw depth and other geometrical features in the<br />

general case.


crack depth, mm<br />

<strong>GMT</strong> Project<br />

<strong>GMT</strong>, <strong>Permissible</strong> <strong>Width</strong> <strong>of</strong> <strong>Front</strong> <strong>Face</strong> <strong>Flaws</strong><br />

20<br />

18<br />

16<br />

14<br />

12<br />

10<br />

8<br />

6<br />

4<br />

2<br />

depth<br />

a/w<br />

Doc # :<br />

Date: 2009-Feb-13<br />

Status: Rev F<br />

Page: 7 <strong>of</strong> 9<br />

An a/w=0.20 crack growing over time, 308 psi<br />

0<br />

0<br />

0.0 50.0 100.0 150.0 200.0 250.0 300.0 350.0 400.0<br />

time, days<br />

Figure 4.1 Flaw Depth and Evolving Aspect Ratio <strong>of</strong> an Initially a/w=0.2<br />

Flaw.<br />

crack depth, mm<br />

1.90<br />

1.85<br />

1.80<br />

1.75<br />

1.70<br />

1.65<br />

1.60<br />

1.55<br />

1.50<br />

1.45<br />

1.40<br />

An a/w=0.20 crack growing over time, 308 psi<br />

Y, a/w variable<br />

Y, a/w =0.2 (fixed)<br />

0.0 100.0 200.0 300.0 400.0<br />

time, days<br />

Figure 4.2 A Comparison <strong>of</strong> Flaw Intensity Geometry Factors from the<br />

Variable Aspect Ratio Case and the Constant Aspect Ratio Case. Since the<br />

flaw tip stress intensity, KI, is proportional to the geometry factor, Y, the<br />

reduced Y in the variable aspect ratio case means a lower KI and hence a<br />

lower crack growth rate.<br />

4.2. Use <strong>of</strong> Flaw <strong>Width</strong> as the Controlling Parameter<br />

0.4<br />

0.35<br />

0.3<br />

0.25<br />

0.2<br />

0.15<br />

0.1<br />

0.05<br />

Aspect Ratio, a/2w


<strong>GMT</strong> Project<br />

<strong>GMT</strong>, <strong>Permissible</strong> <strong>Width</strong> <strong>of</strong> <strong>Front</strong> <strong>Face</strong> <strong>Flaws</strong><br />

Doc # :<br />

Date: 2009-Feb-13<br />

Status: Rev F<br />

Page: 8 <strong>of</strong> 9<br />

Since we are evaluating the permissible size <strong>of</strong> semi-elliptical surface<br />

flaws and since we can non-destructively measure the width but not the<br />

depth <strong>of</strong> the flaws we want to present our results in terms <strong>of</strong> the flaw<br />

width and not depth. It has been noted in section 3.2 that the average<br />

depth <strong>of</strong> front face flaws is 23% <strong>of</strong> the width with a 3σ variation <strong>of</strong> 19%<br />

<strong>of</strong> the width. It has been noted in section 4.1 that it is conservative to<br />

assume that the initial aspect ratio stays constant over time. It will now be<br />

shown that given a measured flaw width, the worst case assumption is that<br />

the depth is 40% <strong>of</strong> the width (rounded up from a mean plus 3σ value <strong>of</strong><br />

39%) rather than 7% <strong>of</strong> the width (the mean minus 3σ aspect ratio). In<br />

other words, it will be shown that the deeper semi-circular flaw is worse<br />

than a shallow surface flaw <strong>of</strong> the same width. Table 4.1 lists the initial<br />

dimensions <strong>of</strong> semi-elliptical surface flaws that grow to failure in one year<br />

at 308 psi. Flaw aspect ratios are held constant as the flaw grows in Table<br />

4.1. The a/w=0.4 flaw has the smallest width so when determining the<br />

acceptable width <strong>of</strong> a flaw we should assume an aspect ratio <strong>of</strong> 0.4.<br />

Note the excellent agreement between Olbert’s results, scaled to 308 psi,<br />

and Cuerden’s. In this revision, Cuerden’s calculation <strong>of</strong> crack growth is<br />

using the more accurate geometry factor used by Olbert. Each <strong>of</strong> us is<br />

using the same crack growth law and are conservatively assuming a<br />

constant flaw aspect ratio. The close agreement therefore indicates that<br />

both computational procedures (Mathcad and Excel) are equivalent.<br />

Olbert’s results at 300 psi were scaled to 308 psi by multiplying the<br />

allowable flaw dimensions by (300/308) 2 since KI=Yσ√a the scaled flaw<br />

depth gives the same KI at 308 psi as the original depth gives at 300 psi.<br />

Cuerden’s calculation at 300 psi gives a flaw depth <strong>of</strong> 6.043 mm<br />

(a/w=0.25) compared to 5.74 mm at 308 psi. Applying the scaling factor<br />

to 6.043 mm gives 5.73 mm which is only <strong>of</strong>f by 0.12%.<br />

Cuerden, 308 psi Olbert, 300 psi Olbert, 308 psi<br />

Design<br />

a/w w a w a w a w a<br />

mm mm mm mm mm mm mm mm<br />

0.4 21.5 8.6 21.5 8.6<br />

0.25 23.0 5.7 24.2 6.1 23.0 5.7 23.0 5.7<br />

0.2 25.1 5.0 26.4 5.3 25.0 5.0 25.0 5.0<br />

0.15 29.5 4.4 31.1 4.7 29.5 4.4 29.5 4.4<br />

0.1 39.1 3.9 41.1 4.1 39.0 3.9 39.0 3.9<br />

Table 4.1 <strong>Permissible</strong> Flaw Sizes for Various Aspect Ratios for One Year <strong>of</strong> Life<br />

at 308 psi Applied Stress. The design value is the smaller <strong>of</strong> the two independent<br />

calculations with Olbert’s result being multiplied by (300/308) 2 to scale the result<br />

to 308 psi. Aspect ratios are fixed as the flaw grows.


5. Results<br />

<strong>GMT</strong> Project<br />

<strong>GMT</strong>, <strong>Permissible</strong> <strong>Width</strong> <strong>of</strong> <strong>Front</strong> <strong>Face</strong> <strong>Flaws</strong><br />

5.1. <strong>Permissible</strong> Initial Flaw at 308 psi for One year life<br />

Doc # :<br />

Date: 2009-Feb-13<br />

Status: Rev F<br />

Page: 9 <strong>of</strong> 9<br />

For an elliptical surface flaw with a/w=0.4, Table 4.1 specifies a<br />

maximum allowed width <strong>of</strong> 21.5 mm at 308 psi for one year <strong>of</strong> exposure.<br />

5.2. <strong>Permissible</strong> Initial Flaw at 231 psi for One year life<br />

[AD1] specifies a maximum front surface stress <strong>of</strong> 154 psi. This was<br />

doubled to provide margin. To assess how conservative this is in terms <strong>of</strong><br />

flaw size, the crack growth calculation has been repeated at 231 psi<br />

(150% <strong>of</strong> the maximum service stress).<br />

The result is a permissible flaw width <strong>of</strong> 36.2 mm which is 68% longer<br />

than the allowable flaw width at 308 psi.<br />

5.3. <strong>Permissible</strong> Initial Flaw at 308 psi for Six Months<br />

6. Conclusions<br />

It has been noted in section 3.2 that the accumulated exposure time is<br />

more likely to be 6 months than one year. Repeating the crack growth<br />

calculation for a six month life gives a permissible flaw width <strong>of</strong> 22.9 mm<br />

which is 7% longer than the allowable flaw width for one year <strong>of</strong> stress<br />

exposure.<br />

<strong>Flaws</strong> <strong>of</strong> up to 21.5 mm wide are permissible on the front face <strong>of</strong> the<br />

<strong>GMT</strong> primary segment. This value has been derived using a conservative<br />

calculation using 3σ crack growth data a 3σ aspect ratio value and twice<br />

the maximum expected stress for twice the estimated duration <strong>of</strong> exposure<br />

to maximum stress. The allowable flaw widths were independently<br />

calculated by Olbert and Cuerden. Small differences between the results<br />

<strong>of</strong> these two calculations were evaluated and traced to slightly different<br />

crack growth laws and flaw stress intensity geometry factors. When the<br />

same crack growth law and geometry factors were used in both<br />

calculations conservatively assuming a constant flaw aspect ration, the<br />

results, allowable flaw width, differ by less than 1%. The results<br />

presented in this report use the most conservative <strong>of</strong> the data and<br />

assumptions initially used by Olbert and Cuerden.<br />

These results apply only to flaws having the same characteristics as those<br />

demonstrating depths <strong>of</strong> less than 40% <strong>of</strong> the width.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!