DOE in TEMPs, T&E Concepts, Test Plans and BLRIPs

1. DOE in TEMPs, T&E Concepts,Test Plans and BLRIPs Lessons Learned from Case Studies

2. Purpose • Discuss lessons learned from past tests • Illustrate how DOE thinking can be applied to TEMPs, Test Plans, and other documents

3. Outline • Overview • Elements of “DOE” Process • Examples • Quantitative, Mission-Oriented Metrics • Coverage of Operational Envelope • Confidence and Power of Test • Summary

4. Overview • Based upon DOT&E initiative: • “Whenever possible, our evaluation of performance must include a rigorous assessment of the confidence level of the test, the power of the test and some measure of how well the test spans the operational envelope of the system.” • IDA conducted analysis of select BLRIPs from last two years • Noted a structured approach to testing that capture many aspects of these concepts • The analysis also identified areas of potential improvement • Modify TEMPs, T&E concept papers, Test Plans, and BLRIPs to incorporate “DOE” concepts

5. Elements of “DOE” Process • Have quantitative, mission-oriented metrics: • What is the question(s) we are trying to answer? • e.g., Can a unit equipped with the Mobile Gun System (MGS) successfully accomplish its missions? • What are the applicable metrics? • Describe how well the operational envelope is covered: • Identify factors that drive performance • e.g., threat, terrain, environment, mission • Identify levels for each factor • Show how well the test covers the operational envelope • For both individual test periods and the overall test program • Calculate the confidence level and power of the test: • Test plan: • Significance, Power, Effect Size, sample size … • Test reports: • XX% confidence intervals • Confidence performance above threshold • Consider whether standard DOE techniques are applicable There is no “one size fits all” solution

6. Quantitative, Mission-Oriented Metrics

7. Mission-Oriented Metrics • Case studies identified several areas for potential improvement • Ensure metrics and KPPs are measureable and testable • As defined, many are not, e.g., “The Mobile Gun System (MGS) primary armament must defeat a standard infantry bunker and create an opening in a double reinforced concrete wall, through which infantry can pass.” • Frequently mission-oriented metrics do not have thresholds • Consider whether they should have a threshold • Is the standard “at least as good as (or better than) the legacy system?” • Do you have quantitative data on the legacy system? Look at metrics during JCIDS process

8. Surveys • Surveys frequently have been qualitative and poorly designed • There is a science behind survey design; use it • Be quantitative (e.g., Likert scale) • During analysis, watch for discrepancies between numerical scores and written comments Be careful with surveys

9. Coverage of Operational Envelope

10. Coverage of the Operational Envelope • 1st Step: Identify factors & levels of interest • In case studies, factors & levels of interest were sometimes specified, other times they were added in retrospective study. • 2nd Step: Determine breadth of coverage of operational envelope • Tools illustrated in following examples: cross-tabular matrices, continuous plots, other graphical representations • These are examples, do not restrict yourself to them • Power analysis can help determine if test design is sufficient • Next section of brief

11. Mobile Gun System (MGS)Coverage of Operational Envelope • 4 Factors: Mission Type, Terrain Type, Threat Level & Illumination • IOT test design builds on evidence from previous events • Mission Rehearsal Exercise prior to unit deployment (basis for Section 231 report) • Field data from unit deployment • IOT scoped to focus on voids in medium and high threat levels Weather: as it occurred; not controlled Key Early deployment changed original DOE plan

12. Mobile Gun System (MGS)Coverage of Operational Envelope Lesson Learned: “DOE” identified gaps in coverage, partially filled from other sources • 4 Factors: Mission Type, Terrain Type, Threat Level & Illumination • IOT test design builds on evidence from previous events • Mission Rehearsal Exercise prior to unit deployment (basis for Section 231 report) • Field data from unit deployment • IOT scoped to focus on voids in medium and high threat levels Weather: as it occurred; not controlled Key Early deployment changed original DOE plan

13. USS VirginiaAnti-Submarine Warfare (ASW) Search • What is the operational envelope? (factors and levels) • Environmental Factors • Shipping Levels and Sea State (ambient noise) • Sound Velocity Profiles (several types – each with different sound propagation characteristics) • Target types and operating modes • SSN (signature, sonar capability/proficiency) • SSK (signature, operating modes, sonar capability/proficiency) • Test submarine configurations (two towed arrays and wide aperture array) • Scenarios (area search, barrier search, cued intercept, multiple targets) • Cross tabular matrix from previous example might not illustrate breadth of coverage appropriately!

14. USS Virginia – ASWCoverage of Operational Envelope • Plot simplifies environmental and target type factors into ordinal comparisons • Only tested Virginia with TB-29 towed array (inadequacy noted in BLRIP) • Area search considered most difficult, other scenarios not examined in IOT&E • Stimulated sensors to simulate multiple target scenario • No SSK testing with Virginia conducted • ARCI data used to provide assessment • Two Virginia tests do not cover entire environmental space

15. USS Virginia – ASWCoverage of Operational Envelope • Plot simplifies environmental and target type factors into ordinal comparisons • Only tested Virginia with TB-29 towed array (inadequacy noted in BLRIP) • Area search considered most difficult, other scenarios not examined in IOT&E • Stimulated sensors to simulate multiple target scenario • No SSK testing with Virginia conducted • ARCI data used to provide assessment • Two Virginia tests do not cover entire environmental space UnknownPerformance Historical Data Sufficient to assess performance Difficult to determine response curve from two SSN tests UnknownPerformance Lesson Learned: “DOE” helped identify gaps

16. USS Virginia – StrikeCoverage of Operational Envelope Strike mission broken into phases with multiple factors/levels PMissile Placement = PEPPAPTGTPLPM

17. USS Virginia – StrikeCoverage of Operational Envelope Strike mission broken into phases with multiple factors/levels PMissile Placement = PEPPAPTGTPLPM Limited missile firings will be discussed later

18. Joint Chemical Agent Detector (JCAD) • What is the operational envelope? (factors and levels) • Agent (9 agents and 2 simulants) • Temperature, water vapor concentration, agent concentration, interferent (continuous) • Environment (sand, sun, wind, rain, snow, fog) • Service (Army, Air Force, Navy, Marine Corps) • JCAD Mode (Monitor, Survey, TIC) • Operator (Any MOS to CBRN Specialist) • TTP (Monitor Mission, Survey Mission, Decon Support)

19. Joint Chemical Agent Detector (JCAD)Coverage of Operational

20. Joint Chemical Agent Detector (JCAD)Coverage of Operational “DOE” applied to full test program for breadth of coverage Response Surface Design applied to chamber tests

21. Joint Chemical Agent Detector (JCAD)DT Chamber Test – Response Surface Design

22. Joint Chemical Agent Detector (JCAD)DT Chamber Test – Response Surface Design “DOE” helps determine whether gaps are significant to the overall assessment

23. Confidence and Power of Test

24. Confidence and Power of Test • Test Planning vs. Test Reporting • Test Planning • What confidence level do we need? • Construct power of test – do we have high probability that the test will detect important differences? • Test Reporting • Provide confidence levels for all results. • Provide confidence above threshold where required.

25. Joint Chemical Agent Detector (JCAD)Power of Test • Power Analysis for JCAD Chamber Test • DT Testing • Statistical Response Surface Design (I-Optimal) • High power test plan *S:N – signal-to-noise ratio, goal detectable difference as a ratio to the design standard deviation

26. Mobile Gun System (MGS)Power of Test • Original Test Plan • (Sample Size = 22) • DOE Interrupted by Deployment • (Sample Size = 16) *S:N – signal-to-noise ratio, goal detectable difference as a ratio to the design standard deviation Lesson Learned: smaller sample size decreases power

27. EA-18G/EA-6B ComparisonConfidence Intervals Figure from DOT&E EA-18G BLRIP Percent Success

28. EA-18G/EA-6B ComparisonConfidence Intervals Figure from DOT&E EA-18G BLRIP Percent Success Confidence intervals make it clear performance is comparable

29. Mobile Gun System (MGS)Confidence Intervals Even without a threshold, confidence intervals quantifies how well the metric was measured

30. MH-60R/S P3IConfidence Above Threshold For both aircraft, all mission failures were due to legacy airframe issues vice P3I systems

31. MH-60R/S P3IConfidence Above Threshold Lesson Learned: Data not available to calculate. Watch data collection and management plan For both aircraft, all mission failures were due to legacy airframe issues vice P3I systems

32. Mobile Gun System (MGS)Data Analysis “DOE” illustrates how performance varies across envelope • Overall Mission Success Rate is 69% • Mission Success tied to unit achieving assigned objectives and unit losses

33. USS Virginia MetricsConfidence Intervals Statistical metrics may require special techniques Provide supplementary details from past testing. Previous Tomahawk testing demonstrated … Use factors and past data to identify limited test scenarios

34. Summary

35. Summary • Next Steps: Modify TEMPs, T&E concept papers, Test Plans, and BLRIPs to incorporate “DOE” concepts • Have quantitative, mission-oriented metrics: • What is the question(s) we are trying to answer? • What are the applicable metrics? • Describe how well the operational envelope is covered: • Identify factors that drive performance • Identify levels for each factor • Show how well the test covers the operational envelope • For both individual test periods and the overall test program • Calculate the confidence level and power of the test: • Test plan: • Significance, Power, Effect Size, sample size … • Test reports: • XX% confidence interval • Confidence performance above threshold • Consider whether standard DOE designs are applicable