Automatic Pipe Routing

Transcription

1 ICCAS2013 in Busan Automatic Pipe Routing To Avoid Air Pockets Kyushu University Dep. of Marine Systems Engineering Yuto Ando, Hajime Kimura

2 Outline Introduction 1. Approach 2. Air Pockets 3. Simulations Conclusion ICCAS 2013 in Busan 1

3 Motivations Oil or Chemical Plants Microchips Ship Building Design of Pipelines ICCAS 2013 in Busan 2

4 Motivations Design of Pipe-Line in Ship Positions of equipment ( valves, etc. ) Piping routes Estimation of safety New Approach Experience of Veterans Design Complete! Al lot of ftime to design. Veterans are decreasing. ICCAS 2013 in Busan 3

5 Motivations Design of Pipe-Line in Ship Positions of equipment ( valves, etc. ) Piping routes Estimation of safety New Approach Experience of Veterans Design Complete! Al lot of ftime to design. Veterans are decreasing. ICCAS 2013 in Busan 3

6 Motivations Design of Pipe-Line in Ship Positions of equipment ( valves, etc. ) Piping routes Estimation of safety New Approach Experience of Veterans Design Complete! Al lot of ftime to design. Veterans are decreasing. ICCAS 2013 in Busan 3

7 Purpose Automatic Pipe-Line Design System PID Automatic Design System Final Design Equipment Placing System Pipe-Routes Equipment Layout Routing System ICCAS 2013 in Busan 4

8 Purpose Connector (Movable) Automatic Pipe-Line Design System Pump (Immovable) PID Valve (Movable) Automatic Design System Final Design Equipment Placing System Pipe-Routes Equipment Layout Routing System ICCAS 2013 in Busan 4

9 Outline Introduction 1. Approach 2. Air Pockets 3. Simulations Conclusion ICCAS 2013 in Busan 5

10 Approach of Routing System Routing of pipelines Routing in a network ICCAS 2013 in Busan 6

11 Approach of Routing System ICCAS 2013 in Busan 7

12 Approach of Routing System Directed and Weighted Graph ICCAS 2013 in Busan 7

13 Approach of Routing System Dijkstra s Method Directed and Weighted Graph ICCAS 2013 in Busan 7

14 Approach of Routing System The Optimized Pipe Route Dijkstra s Method Directed and Weighted Graph ICCAS 2013 in Busan 7

15 Pipe Pieces and Items Pipe Pieces Straight Elbow Bending Part ICCAS 2013 in Busan 8

16 Pipe Pieces and Items Pipe Pieces In the graph Current node Next node Edge Weight of the Edge = (Manhattan distance + Costs of the Pipe Piece) x Diameter ICCAS 2013 in Busan 9

17 Pipe Pieces and Items Pipe Pieces In the graph Current node Next node Weight of the Edge = (Manhattan distance + Costs of the Pipe Piece) x Diameter ICCAS 2013 in Busan 9

18 Design Objectives Obstacle The shortest path on the network Pipe Rack Short Aisle Not winding Avoid aisles Set on pipe racks ICCAS 2013 in Busan 10

19 Design Objectives Obstacle The shortest path on the network Pipe Rack Short Aisle Not winding Set on pipe racks Avoid aisles Change the weight of edges ICCAS 2013 in Busan 10

20 Design Objectives Obstacle Pipe Rack Aisle : Light weighted edge ( = Pipe-rack k) : Heavy weighted edge ( = Aisle ) ICCAS 2013 in Busan 10

21 Design Objectives Minimize the total length of piping routes Minimize the number of elbows and bending parts Avoid aisles as much as possible Pass through pipe-rack areas as much as possible ICCAS 2013 in Busan 11

22 Design Objectives Minimize the total length of piping routes Minimize the number of elbows and bending parts Avoid aisles as much as possible Pass through pipe-rack areas as much as possible ICCAS 2013 in Busan 11

23 Design Objectives Minimize the total length of piping routes Minimize the number of elbows and bending parts Avoid aisles as much as possible Pass through pipe-rack areas as much as possible ICCAS 2013 in Busan 11

24 Design Objectives Minimize the total length of piping routes Minimize the number of elbows and bending parts Avoid aisles as much as possible Pass through pipe-racks areas as much as possible ICCAS 2013 in Busan 11

25 Design Objectives Minimize the total length of piping routes Minimize the number of elbows and bending parts Avoid aisles as much as possible Pass through pipe-racks areas as much as possible = The shortest path in the weighted graph ICCAS 2013 in Busan 11

26 Order of Routing Order of routing in the system = From the largest to the smallest Obstacle G G Pipe B S G Pipe C S S Obstacle ICCAS 2013 in Busan 12

27 Order of Routing Order of routing in the system = From the largest to the smallest Obstacle G G Pipe B S Pipe A G Pipe C S S Obstacle ICCAS 2013 in Busan 12

28 Order of Routing Order of routing in the system = From the largest to the smallest Obstacle G G Pipe B S Pipe A G Pipe C S S Obstacle ICCAS 2013 in Busan 12

29 Order of Routing Order of routing in the system = From the largest to the smallest Obstacle G G Pipe B S Pipe A G Pipe C S S Obstacle ICCAS 2013 in Busan 12

30 Order of Routing From the largest diameter to the smallest diameter Random choice from pipes with same diameters Random choice Random choice ICCAS 2013 in Busan 13

31 Outline Introduction 1. Approach 2. Air Pockets 3. Simulations Conclusion ICCAS 2013 in Busan 14

32 What s Pocket? U-shaped pipe piece in a vertical direction Liquid or gas settle at Pocket z Drain traps Not preferred pipe piece Pump Engine Drain Traps ICCAS 2013 in Busan 15

33 What s Pocket? U-shaped pipe piece in a vertical direction Liquid or gas settle at Pocket z Drain traps Undesirable pipe piece Pump Engine Drain Traps ICCAS 2013 in Busan 15

34 Algorithm to avoid Pockets Loading of Start and Goal point Method1 (Restriction Method) Found! Route without pockets Not Found : At least one pocket in the route Method2 (Penalty Method) Not Found No pipe pp route Found! Route avoiding pockets as much as possible ICCAS 2013 in Busan 16

35 Algorithm to avoid Pockets Loading of Start and Goal point Method1 (Restriction Method) Found! Route without pockets Not Found : At least one pocket in the route Method2 (Penalty Method) Not Found No pipe pp route Found! Route avoiding pockets as much as possible ICCAS 2013 in Busan 16

36 Method1: Restriction Method Restriction Method - Delete candidates making U-turn - Reduce size of network No pocket in the route ICCAS 2013 in Busan 17

37 Method1: Restriction Method z Candidates with the highest costs ICCAS 2013 in Busan 18

38 Method1: Restriction Method z Candidates with the highest costs Route without Pockets Remove from candidates ICCAS 2013 in Busan 18

39 Simulation of Restriction Method Solution by Old System (Without consideration of Pockets) Solution by Restriction Method ICCAS 2013 in Busan 19

40 Simulation of Restriction Method Other test case Complex route ( Route with at least one Pocket) No solution by Restriction Method Pocket is necessary ICCAS 2013 in Busan 20

41 Algorithm to avoid Pockets Loading of Start and Goal point Method1 (Restriction Method) Found! Route without pockets Not Found : At least one pocket in the route Method2 (Penalty Method) Not Found No pipe pp route Found! Route avoiding pockets as much as possible ICCAS 2013 in Busan 21

42 Method2 : Penalty Method Penalty Method - Add penalties on edges connecting vertically. - Search horizontal candidates as priority. Route involves Pockets as few as possible. Add Penalty ICCAS 2013 in Busan 22

43 Method2 : Penalty Method z Start Goal Red arrows : With penalties Blue arrows : Normal Candidates with the highest costs Avoid moving vertically as much as possible Increase of search time Route avoiding Pockets as much as possible ICCAS 2013 in Busan 23

44 Method2 : Penalty Method z Start Goal Red arrows : With penalties Blue arrows : Normal Candidates with the highest costs Avoid moving vertically as much as possible Increase of search time Route avoiding Pockets as much as possible ICCAS 2013 in Busan 23

45 Method2 : Penalty Method z Start Goal Red arrows : With penalties Blue arrows : Normal Candidates with the highest costs Avoid moving vertically as much as possible Increase of search time Route avoiding Pockets as much as possible ICCAS 2013 in Busan 23

46 Simulation of Penalty Method Solution by old System Solution by Penalty Method ICCAS 2013 in Busan 24

47 Outline Introduction 1. Approach 2. Air Pockets 3. Simulations Conclusion ICCAS 2013 in Busan 25

48 Simulations Purpose of simulations Comparisons of old and new Systems, different mesh sizes, and orders of routing Test model Design Space : 6 x 6 x 6 [m] Pipes : 0.8[m] x 1, 0.6[m] x 2, 0.4[m] x 4, 0.3[m] x 6 Order : From the largest pipe Cost of Elbow : 0.1 Cost of Bending : 0.3 Penalty of vertical edge : x2 ICCAS 2013 in Busan 26

49 Simulations Purpose of simulations Comparisons of old and new Systems, different mesh sizes, and orders of routing Test model Design Space : 6 x 6 x 6 [m] Pipes : φ 0.8[m] x 1, φ 0.6[m] x 2, φ 0.4[m] x 4, φ 0.3[m] x 6 Order : From the largest pipe Cost of Elbow : 0.1 Cost of Bending : 0.3 Penalty of vertical edge : x2 ICCAS 2013 in Busan 26

50 Simulation1 (Comparison of Old and New Systems) Pocket 90 minutes 70 minutes Old System New System (Without consideration of Pockets) (With Method 1 and 2) Succeeded to avoid making Pockets ICCAS 2013 in Busan 27

51 Simulation2 ( Comparison of different orders ) 1 (φ 0.9m) 1 (φ 0.9m) 3 (φ 0.6m) 2 (φ 06 ) 2 (φ 0.6m) 2 (φ 0.6m) 3 (φ 0.6m) Order of routing: Strong influence to the final design. ICCAS 2013 in Busan 28

52 Simulation3 (Comparison of mesh sizes) Mesh: 0.15 [m] Time: 70 [minutes] Mesh: 0.25 [m] Time: 30 [minutes] Small Mesh Size Accurate Routing Long Search Time Large Mesh Size Rough Routing Short Search Time Mesh Size: Strong influence to the routes. ICCAS 2013 in Busan 29

53 New Mesh Dividing Method Regular meshes only ICCAS 2013 in Busan 30

54 New Mesh Dividing Method Regular meshes only : Regular Lines : Additional Lines Around obstacles, pipes, aisles In pipe-racks Regular and additional meshes On start and goal points ICCAS 2013 in Busan 30

55 New Mesh Dividing Method Regular and additional meshes Regular meshes only In short time In long time Increase of items ( ex. Obstacles ) = Increase of additional meshes ICCAS 2013 in Busan 31

56 New Mesh Dividing Method Regular Mesh System Additional Mesh System Num. of Meshes in Regular Mesh << Num. of Meshes in Add. Mesh Increase of search time!! ICCAS 2013 in Busan 32

57 Outline Introduction 1. Approach 2. Air Pockets 3. Simulations Conclusion ICCAS 2013 in Busan 33

58 Conclusion New automatic pipe routing system Features Minimize total length, Minimize number of bending, Pass through pipe-racks, Avoid aisles, Avoid making Pockets - Restriction Method ( Reduce the size of graph) - Penalty Method (Add penalty to the vertical edges) ICCAS 2013 in Busan 34

59 Conclusion New automatic pipe routing system Features Minimize total length, Minimize number of bending, Pass through pipe-racks, Avoid aisles, Avoid making Pockets - Restriction Method ( Reduce the size of graph) - Penalty Method ( Add penalties to the vertical edges ) ICCAS 2013 in Busan 35

60 Future Tasks Order of Routing Search Time More practical routes Ship Hull Equipment Placing System Large Curved Pipe This routing system will be opened for free at u.ac.jp/gen/index.html Thank you! ICCAS 2013 in Busan 36