Modeling and Analysis of Integration Processes for Engineering Systems

Transcription

1 Modeling and Analysis of Integration Processes for Engineering Systems Meir Tahan and Joseph Z. Ben-Asher (24) Student 8 Jon Gibbs Page 1

2 Types of Integration Single Stage Small number of subsystems Assumes subsystems acquired simultaneously Incremental System too complex for single stage Subsystems can arrive independently Multiple methods Typically have associated time and cost benefits Page 2

3 Large Scale System Development Issues Subsystem Development Synchronizing Tasks System Development Integration of Subsystems System Performance Verification Page 3

4 Relationship Strengths 4 Types of Flow: Information, Energy, Material, Physical Degree : No significant relationship Degree 1: One type of flow only Degree 2: Significant flow of two types Degree 3: Significant flow of three types Sharman and Yassine [24] Page 4

5 Case Study - UAV Four UAV subsystems 1. Engine 2. Flight control 3. Optronic payload 4. Payload control How should these systems be integrated? Page 5

6 Single Stage Integration Single-stage integration Act1 Act2 Act3 Act4 7 d*int12 8 d*int34 9 Int Subsystems prepared for integration Tightly coupled subsystems integrated separately Final systems integration of integrated subsystems Total Time: Max(t 1, t 2, t 3, t 4 ) + d*t 1,2 + d*t 3,4 + t 1,2,3,4 Image by MIT OpenCourseWare. Adapted from Figure 2 on p. 65 in Tahan, Meir, and Joseph Z. Ben-Asher. Modeling and Analysis of Integration Processes for Engineering Systems. Systems Engineering 8, no. 1 (25): Page 6

7 Incremental Integration Incremental integration. Act1 Act2 Act3 Act d*int12 q,int34 S 9 1 Int1234 p,int12 7 d*int34 q=1-p Image by MIT OpenCourseWare. Adapted from Figure 3 on p. 65 in Tahan, Meir, and Joseph Z. Ben-Asher. Modeling and Analysis of Integration Processes for Engineering Systems. Systems Engineering 8, no. 1 (25): Subsystems prepared for integration First set of tightly coupled subsystems ready for integration triggers integration start Assumes that remaining two subsystems ready after first set is integrated Integration of all subsystems then performed Total Time: Min of [Max(t 1,t 2 ), Max(t 3,t 4 )] + t 3,4 + d*t 1,2 + t 1,2,3,4 Page 7

8 Success Probability Summary of the probabilities of sucess in meeting the time requirement Case study Case 1 Case 2 d Single stage Incremental Image by MIT OpenCourseWare. Adapted from Table 2 on p. 68 in Tahan, Meir, and Joseph Z. Ben-Asher. Modeling and Analysis of Integration Processes for Engineering Systems. Systems Engineering 8, no. 1 (25): Incremental integration has higher probability of success Single stage integration will perform poorly with complex systems Page 8

9 Types of Incremental Integration Option A: Single Stage T s =Max(t 1,t 2,t 3,t 4 ) Option B: Integrate first available units: T s =Min[Max(t 1,t 2 ),Max(t 3,t 4 )] Then integrate other units in order of arrival Option C: Integration in pairs followed by full system integration Minimize Total Integration Time t1 t2 t3 t4 5 6 q,t34 d*t12 S 9 1 t1234 p,t12 7 d*t34 Image by MIT OpenCourseWare. Adapted from Figure 4 on p. 66 in Tahan, Meir, and Joseph Z. Ben-Asher. Modeling and Analysis of Integration Processes for Engineering Systems. Systems Engineering 8, no. 1 (25): Option D: Integrate the first pair then integrate all available units Page 9

10 Incremental Options C & D Option C (1) Integrate first available Min[Max (t 1,t 2 ),Max (t 3,t 4 )] (2) Integrate in pairs (t 1,2,t 3,4 ), (3) Full systems integration (t 1,2,3,4 ) T tot =Min[Max (t 1,t 2 ),Max (t 3,t 4 )]+t 3,4 +t 2,3 +t 1,2,3,4 Option D (1) Assume a simulator for unit 1 Min[t 2,Max (t 3, t 4 )] (2) Integrate remaining pairs with some difficulty d*t 3,4 or d*t 1,2 (3) Full systems integration T tot =Min[t 2,(Max (t 3,t 4 )+t 1,2 ]+d*t 3,4 +t 2,3 +t 1,2,3,4 Page 1

11 Results.1 Difficulty Factor mainly determines break even point Tot w B D Image by MIT OpenCourseWare. Adapted from Figure 1 on p. 73 in Tahan, Meir, and Joseph Z. Ben-Asher. Modeling and Analysis of Integration Processes for Engineering Systems. Systems Engineering 8, no. 1 (25): P success [%] 2 1 Option B Option D X[%] Replacement time of simulator can be traded with difficulty factor to decide which method has the fastest expected time Many firms know the replacement time of a simulator with more certainty than component completion times Image by MIT OpenCourseWare. Adapted from Figure 11 on p. 74 in Tahan, Meir, and Joseph Z. Ben-Asher. Modeling and Analysis of Integration Processes for Engineering Systems. Systems Engineering 8, no. 1 (25): Page 11

12 Decision Considerations Single Vs. Incremental Consider level of complexity Simulator: Consider replacement time and level of uncertainty Cost of test equipment and labor Type of incremental integration Knowledge of PDF for each method Page 12

13 Very Light Jet Case Study Is it faster to prototype or not? Approach 1) Obtain FAA approval first, make changes, tackle manufacturing learning curve after Approach 2) Prototype first and tackle manufacturing learning curve, obtain FAA approval second, make certification changes last Three Variables: T lc,t faa,d*t rework Page 13

14 Conclusions Incremental integration almost always requires less time than single stage integration Incremental integration may be required for complex systems Multiple levels of incremental integration exist Page 14

15 Integration Example Boeing 787/GEnX Six Primary Modules Fan, LPC, HPC, Combustor, HPT, LPT Image by MIT OpenCourseWare. Page 15

16 Boeing 787/GEnX Integration Aircraft and engines are typically developed in parallel at separate companies and with minimal final integration (relative to aircraft/engine design). What allows this incremental integration to occur? Aircraft engine design is typically divided into modules (subsystems). One could argue that the relationship between modules is both tightly and weakly coupled. Why? Page 16

17 Discussion Questions Many of the incremental integration models do not assume a particular path, but rather perform sub-system and system integration when the inputs are available. Is there a benefit from planning a target path for integration? The paper presents a decent back-to-back study of single stage and incremental integration through various modeling techniques. Is there a benefit for a program to evaluate the integration methodology using these techniques? If so, would the result be quantitative or just qualitative? What level of simulator accuracy is required for incremental integration? Page 17

18 MIT OpenCourseWare Fundamentals of Systems Engineering Fall 29 For information about citing these materials or our Terms of Use, visit: