NGAO: Cost Comparison with First Light TMT AO

advertisement

NGAO:

Cost Comparison with First Light TMT AO

Peter Wizinowich, Richard Dekany, Don Gavel with input & review by Brent Ellerbroek

SSC Meeting

November 3, 2008

Presentation Sequence

• SDR Reviewer comments leading to this cost comparison

• NGAO SDR Cost Estimate & Methodology

• Cost Comparison Goal, Products & Approach

• High-level Summary of Results

• Cost Comparison

• Conclusion

Conclusion Preview:

We & TMT agree that NGAO is traceably less expensive than the 1 st light TMT AO system & we understand why

- Larger NGAO contingency needed in some areas

- Quote needed for laser & will be obtained soon

2

SDR Reviewer Comments

• “Based on the cost and schedule of past and planned projects of lower or similar complexity, the review panel believes that the NGAO project cost and schedule are not reliable and may not be realistic.

Contingencies are also too tight. In particular, the time of 18 months allocated for manufacturing and assembly and 6 months for integration and test, is probably optimistic by a large amount.”

• Relevant to this point they also said:

– “The review panel believes that Keck Observatory has assembled an

NGAO team with the necessary past experience … needed to develop the

Next Generation Adaptive Optics facility for Keck.”

– “The proposed schedule and budget estimate have been carried out with sound methodology”

• Clarification: Reviewers thought our lab and telescope I&T durations were smaller by 2x than our plan (they are 6 & 12 months, respectively).

3

Response to SDR Reviewer Comments

• “We will develop another level of cost estimation during preliminary design.”

– During system design, we considered project costs composed of hundreds of tasks and specific procurements

– Reviewers noted “the budget book showed some pages not complete or no estimates for the words in the task”. Our choice to produce cost sheets for each WBS in each of 4 project phases resulted in blank sheets

• e.g., no subsystem work during the delivery & commissioning phase

– During PD, we will revise the basis of estimate to move more items from engineering judgment and analogy to direct vendor quote basis

– Reviewers felt that the laser procurement needs more time & should start much earlier. We agree.

• Requested next level of cost information for TMT NFIRAOS project to better understand the reasons for the cost estimate differences

– This brings us to this presentation …

4

Cost Estimation Methodology

(KAON 546)

• Cost estimation spreadsheets

– Based on TMT Cost Book approach, simplified for SD phase

– Prepared for each WBS element (~75 in all)

– Prepared for each of 4 phases

• Preliminary design, detailed design, full scale development, delivery/commissioning

– Prepared by technical experts responsible for deliverables

– Process captures

• WBS dictionary

• Major deliverables

• Estimates of labor hours

• Estimates of non-labor dollars (incl. tax & shipping) & travel dollars

• Basis of estimate (e.g. vendor quote, CER, engineering judgment)

• Contingency risk factors & estimates

• Descope options

– Standard labor classes, labor rates & travel costs used

5

Cost Comparison Goal, Products & Approach

• Goal of Comparison:

– Increase confidence in the NGAO cost estimate

• Products

– Updated cost estimation data base to take advantage of comparison information

– Cost comparison presentation (this presentation)

• Approach

– Identify differences between systems

– Identify differences in assumptions

– Identify similarities & differences in effort & procurement estimates, & resolve/justify these differences

– As appropriate update the NGAO cost estimate

– Identify other support for NGAO estimates

– Respond to reviewer sense that “methodology good, but not fully executed”

• “NFIRAOS” used as shorthand for the full TMT 1 st light AO configuration

– Updates to TMT cost estimate since 2006 (globally < 5%) not incorporated

6

Cost Comparisons

(in FY08 $ as reported at the SDR)

NGAO Cost Estimate Comparisons

100

90

80

70

1 st Gen 30m AO

60

50

40

2 nd Gen 8-10m AO

30

20

10

0

1 st Gen 8-10m AO

Upgrades

N

G

W

FC

P

AL

M

30

00

K

2

LG

S

G

P

I

G

E

M

S

N

G

AO

N

FI

R

A

O

S

7

Major Differences between NFIRAOS & NGAO

• 30 m versus 10 m

– ~ 3x physical size of AO system

– 3x smaller image  more sensitive to image motion & vibrations

– 3x more perspective elongation  more laser power for same performance

&/or better detectors

– 9x LGS focus change with zenith angle

(1.4 m from zenith to 65

)

– Physical paths for laser beam transport much longer

• New versus existing telescope

– Interfaces and observatory software not as well known for TMT

– TMT, NFIRAOS & Instrument I&T could be overlapping

– 2 nd (or 3 rd ) Keck AO system versus 1 st TMT AO system

– Need to be conservative for TMT first light system

• Architecture & Technologies

– Different lasers

(new laser suppliers have emerged since TMT 2006 CoDR)

– MCAO versus MOAO

(2 large high order DMs vs 1 lower order DM + MEMS; & different I&T issues)

– NFIRAOS cooled to -30 

C versus -15

C for NGAO

8

Major Differences between NFIRAOS & NGAO

NGAO (below) roughly to NFIRAOS (left) scale

9 x 4 x 3 m

(blue) box

3.8 m

9

High-level Results of Cost Comparison

• NGAO $42.2M SDR estimate is 48% of NFIRAOS $88.2M

CoDR estimate

– NGAO $34.5M without contingency is 52% of NFIRAOS $66.7M

– All in FY08 $

• Cost differences attributed to the following major factors:

Largest items in $46M difference

Category

Contingency

$M

13.8

Laser Procurement

Labor Rates

Component Development

DM & Tip/Tilt Procurement

9.4

6.5

4.7

4.5

RTC System

LGS WFS

+/- other items

3.9

2.7

Total = 45.5

Conclusions:

Modest contingency increase likely needed

Need quote for new laser

No significant adjustments required - differences largely understood

10

Cost Category Comparison

Labor

Category

NFIRAOS

NGAO hours

203164

$k

20535

231944 16045

NFIRAOS-NGAO -28781 4490

NFIRAOS/NGAO 0.88

1.28

Non-Labor

$k

45151

16804

28347

2.69

# Trips

380

549

-169

0.69

Travel

$k

972

1681

-709

0.58

Sub-total Contingency

$k $k %

66660 21530 32%

34530 7697 22%

32130 13834 10%

1.93

2.80

1.45

• All costs in FY08 dollars

– NFIRAOS costs have been inflated by 4%/year for 2 years

• NFIRAOS cost estimate is 2.1x NGAO

– $32.1M difference without contingency ($46.0M with contingency)

• Major difference is in non-labor (e.g., procurements)

– Factor of 2.7 or $28.3M, but less TMT work is performed “in-house”.

• Other significant differences:

– Factor of 1.3 or $4.5M in labor even though NGAO has 14% more labor

– NFIRAOS uses 32% contingency versus 22% for NGAO

Total

$k

88191

42227

45964

2.09

11

Labor Rate Comparison

Labor

Category

NFIRAOS

NGAO hours

203164

$k

20535

231944 16045

NFIRAOS-NGAO -28781 4490

NFIRAOS/NGAO 0.88

1.28

• NFIRAOS average rate = 1.46x NGAO

– NFIRAOS average rate = $101/hr

– NGAO average rate = $69/hr

– Benefit rates are the same

– 24% burden rate versus 19% at CIT & 0% at WMKO & UCO

– NGAO uses actual CIT, UCO & WMKO rates

• 1.46x represents $6.5M of the $20.5M of NFIRAOS labor

12

> $1M

WBS Category Comparison

NFIRAOS -

Category

Management

System Eng

AO System

RTC System

Laser System

Laser

Operation Tools

I&T

Non-Labor ($k) NFIRAOS -

NFIRAOS

0

0

10667

11669

2474

15575

30

58

Subtotal = 40473

Component Dev 4678

NGAO

354

265

4757

2551

1554

5691

39

791

16001

0

NGAO

-354

-265

5910

9119

920

9884

-9

-733

24472

4678

Facility Mods 0

Total = 45151

803

16804

-803

28347

NFIRAOS /

NGAO

0.0

0.0

2.2

4.6

1.6

2.7

0.8

0.1

2.5

0.0

2.69

Labor (hrs)

NFIRAOS NGAO

23979 42310

31538

62924

6396

43988

25506

56527

16819

15041

0

15975

3440

18295

18364 37050

203164 214987

0

0

0

16957

203164 231944

NGAO

-18331

6032

6398

-10423

28947

-3440

-2320

-18686

-11824

0

-16957

-40604

> 5 wy

NFIRAOS /

NGAO

0.6

1.2

1.1

0.4

2.9

0.0

0.9

0.5

0.9

0.0

0.88

• Start by removing Component Development & Facility Modifications

– NFIRAOS investment (not needed for NGAO) in polar coordinate CCDs & readout electronics to address perspective elongation

– NFIRAOS reserves $0.5M for NIR sensor development (NGAO to use available detectors)

– NGAO modifications to telescope facility (not needed for NFIRAOS) for new

AO system & laser, & to remove old systems

13

AO System Comparison

AO Components

LGS WFS

Low order WFS

Equipment

Source simulators

Optics

Acquisition camera

NGS WFS / TWFS

Support structure

TT mitigation

Device control

Enclosure

Total =

Non-labor ($k) NFIRAOS

NFIRAOS NGAO

3721 1618

- NGAO

2102

NFIRAOS

/ NGAO

2.3

2.8

2785

1142

896

1324

223

247

107

0

1007 1778

1 1141

135 761

624 699

69 154

237 9

113

52

-6

-52

6.6

2.1

3.2

1.0

0.9

116

107

10667

282 -166

618 -511

4757 5910

0.4

0.2

2.2

• Sorted by non-labor cost difference

Labor (hrs) NFIRAOS

NFIRAOS NGAO

16200 7450

- NGAO

8750

NFIRAOS

/ NGAO

2.2

0

5693

3488

9161

1690

5042

9520

1695

1865

14340

578

3384

-9520

3998

1623

-5179

1113

1658

0.0

3.4

1.9

0.6

2.9

1.5

2.1

4050

0

11960

5640

62924

1920 2130

3180 -3180

12075 -115

520 5120

56527 6398

1.0

10.8

1.1

• Cost of many items, especially LGS WFS & Optics, impacted by different scales of TMT & Keck

• Equipment & source simulators represent different philosophies

– NFIRAOS equipment includes high resolution test WFS, jigs & fixtures, & a turbulence generator versus NGAO alignment tools

14

LGS WFS Assemblies

Each lens is ~ 0.4m in diameter

Six zoom assemblies;

18 mechanisms

Entire ass’ly is

~ 0.5 x 0.5 m

Category

Non-labor (FY08)

Labor

Package size

WFS channels

Lenslet arrays

CCD

Camera

Mechanisms

Travel Range

Heritage

NFIRAOS

$3.7M

16,200 hrs

~ 3 m

6 zoom-optics channels with separating periscopes

30 mm diameter

Custom radial-format CCD

Custom controller

18

130 - 590 mm

Significant new optical challenges

NGAO

$1.6M

7,450 hrs

~ 0.5 m

9 optically simple channels

4 mm diameter

CCID56 (modest advance)

Commercial SciMeasure camera

45

1 to 150 mm

Considerable Keck AO heritage

Note: NFIRAOS PDR design corrects each WFS individually

 > ~ $1M savings in lens costs compared with aspheric CoDR design

15

Category

Correctors

Wavefront Correctors Comparison

Labor (hrs)

NFIRAOS NGAO

3620 3040

NFIRAOS - NFIRAOS

NGAO

580

/ NGAO

1.2

NFIRAOS

61x61 DM

73x73 DM

Cabling

Amps & HV

TT platform

Total =

Non-labor

($k) NGAO

1714 20x20 DM

1943 MEMS64

53 4x MEMS32

1298 DM TT platform

952 MEMS TT platform

5960 Total =

Non-labor

($k)

515

412

309

206

41

1483

• MCAO vs MOAO

• NFIRAOS DMs need to be developed

• NGAO DMs are commercially available

– ROM for MEMS64 (being developed for GPI)

• Estimates based on quotes

18

Category

RTC

RTC Comparison

Labor (hrs) NFIRAOS -

NFIRAOS NGAO

3130 13779

NGAO

-10649

NFIRAOS

/ NGAO

0.23

NFIRAOS

RTC design

RTC construct

PSF RT boards

RTC disk array

Total =

Non-labor

($k) NGAO

2163 RTC vendor

3245 FPGAs

54 Interface boards

248 Equipment

5710 Total =

Non-labor

($k)

480

375

168

45

1067

• Similar complexity & parallel processing approaches

• NFIRAOS assumes $5.4M RTC contract based on quote

– Includes ~ 6.9 work-years at industrial rates

• NGAO assumes mostly in-house development

– Includes 6 extra work-years of labor vs NFIRAOS

– NGAO hardware is based on quotes

• Need to add PSF RT boards ($100k should meet NGAO needs)

– Vendor quotes to be obtained prior to PDR to better anchor our estimate

• Less complex systems for comparison:

– NGWFC $2M, including $0.72M Microgate (3 RTC systems) subcontract

– Gemini MCAO RTC subcontract was $0.88M

– GPI $0.65M; PALM-3000 $0.60M

19

Laser Procurement Comparison

Laser

Procurement

Laser

Non-labor ($k) NFIRAOS

NFIRAOS NGAO

15575 5691

- NGAO

9884

NFIRAOS

/ NGAO

2.7

Labor (hrs) NFIRAOS

NFIRAOS NGAO

0 3440

- NGAO

-3440

NFIRAOS

/ NGAO

0.0

• Laser Differences

– NFIRAOS assumes 3x 50W LMCT lasers quote

• $6M design & $8.4M construction in FY06

• Reduced by ~ $1-2M at PDR by assuming fixed gravity vector

– NGAO assumes 2x 50W SOR-type lasers

• NGAO estimate based on our previous experience trying to set up a company to build SOR-type lasers for Gemini & Keck

– FASORtronics recently set up to commercialize this laser

• We are participating in ESO’s call for laser preliminary designs

– ROMs will be available within a month

– A fixed price quote will be available at PDR (Aug/09)

• The SDR reviewers questioned whether 100W would be adequate for the highest order correction

– After re-assessment we believe that 100W has sufficient margin by a factor of at least 1.5

20

Laser System Comparison

Laser System

Components

Enclosure

Launch facility

Safety systems

System control

Total =

Non-labor ($k) NFIRAOS NFIRAOS

NFIRAOS NGAO

699

1154

224

- NGAO

51 648

1340 -187

36 188

397

2474

126 271

1554 920

/ NGAO

13.6

0.9

6.2

3.1

1.6

Labor (hrs)

NFIRAOS

760

20496

5384

NGAO

NFIRAOS

- NGAO

1224 -464

3900 16596

1812 3572

NFIRAOS

/ NGAO

0.6

5.3

3.0

17348

43988

8105 9243

15041 28947

2.1

2.9

• No major procurement differences ( > $1M)

– TMT system physically larger

• Biggest difference is the launch facility labor

– NGAO labor based on K1 LGS launch facility

– NGAO asterism generator based on LGS WFS pick-off mechanisms

• NFIRAOS estimate may be somewhat conservative (CoDR level)

• NGAO may be slightly optimistic

– Not obviously missing anything after comparison to NFIRAOS

– At most could imagine adding 2 work-years to estimate

21

I&T Comparison

I&T Non-labor

AO lab I&T

Laser lab I&T

Laser telescope I&T

AO telescope I&T

Total =

Non-labor ($k) NFIRAOS NFIRAOS

NFIRAOS NGAO

0 113

- NGAO

-113

17

0

45

72

40

51

-55

-40

-6

62 276 -214

/ NGAO

0.0

0.2

0.0

0.9

0.2

I&T Labor (hrs) NFIRAOS NFIRAOS

NFIRAOS NGAO

5526 8480

- NGAO

-2954

4664

2528

5646

3126

4838

19160

1538

-2310

-13514

18364 35604 -17240

/ NGAO

0.7

1.5

0.5

0.3

0.5

• NGAO has more labor in all but one category than NFIRAOS

– NFIRAOS laser lab I&T includes a full off-telescope system test

– NFIRAOS tel. I&T only covers to beginning of science commissioning

• ~ 25% of NGAO AO telescope I&T is for science commissioning

• NGAO estimate takes into account past experience with a detailed breakdown

• Possibility that we are not comparing apples to apples

– NFIRAOS may cover some under system eng. & I&T contingency

– NGAO assumes that each subsystem is complete & has met its requirements prior to lab I&T (presumably also true for NFIRAOS)

22

Contingency

Comparison

Category

Management

System Eng

Contingency %

NFIRAOS NGAO

18% 7%

$7.7M NGAO contingency

AO System

RTC System

Laser System

27%

36%

46%

28%

$22.0M NFIRAOS contingency

• Management & System Engineering:

Laser

I&T

Component Dev

Facility Mods

26%

32%

26%

0%

Total = 32%

16%

31%

24%

26%

19%

22%

0%

23%

22%

– 22% more NGAO labor pre-contingency

– Both NGAO & NFIRAOS assume level of effort

– Keck interfaces & telescope well known + 2 nd generation AO system

• AO & Laser systems: Contingencies fairly close

• RTC system: NFIRAOS at pre-CoDR study estimate, but have vendor quote

– 46% NGAO RTC contingency would require $300k more.

• Laser: NGAO 19% contingency too low. 50% would require $2M more.

• I&T: NGAO plan has ~1.7x more labor excluding science commissioning

• Contingencies generated using the same risk evaluation methodology

– Will evaluate whether applied adequately & consistently for NGAO

NFIRAOS

/ NGAO

2.5

1.7

1.2

1.9

1.1

1.4

1.4

0.0

1.45

23

Conclusions

• NGAO is traceably less expensive than NFIRAOS & we understand why

• Some areas identified that require more work:

– Contingency rates need to be re-evaluated

• At minimum should be increased for laser & potentially for RTC

– Laser procurement estimate needs to be more solidly based

• Will have ROMs soon & a fixed price quote for PDR through ESO collaboration

– Minor items: Laser system labor & cost of RTC labor

• NFIRAOS comparison was worthwhile for determining confidence in NGAO estimate.

– Methodology largely gave us reasonable system design estimates

• A cost review will be part of the NGAO PDR

– We will include NFIRAOS PDR estimate comparison

24

Download