Professional Documents
Culture Documents
www.elsevier.com/locate/eswa
Abstract
This study presents an automatic pipe-routing algorithm accommodating all major detail-design facets. First, the algorithm uses patternmatch methods to provide candidate paths. A cell-generation method is developed which satisfies geometric constraints. This makes the
generation and evaluation of paths effective and efficient. Next, various non-geometric aspects, such as material costs, installation costs, and
valve operability, are assessed from a fiscal point of view. Then, from a tree of combinations, the algorithm chooses an appropriate path for
each pipeline from the candidate paths. Finally, a general approach toward detail design automation is suggested. The software
implementation was done with Microsoft Visual Basic 6.0 and Access 2000, Heide Corporation Intent! for AutoCAD 2000, and AutoDesk
AutoCAD 2000. q 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Pipe-routing; Design automation; Expert system; Cell generation method
1. Introduction
Piping-system design for ships usually has five consecutive phases: preliminary design, functional design, detail
design, production engineering, and system-support information (ISO/CDC, 1996). Pipe routing is the most important
activity during the detail-design phase because it takes over
50% of the total detail-design man-hours and all other
activities of detail design depend on it. Since detail design is
characteristically less creative and more routine than the
earlier steps in the design process (Kang, Myung, & Han,
1999), a computer program that has an automatic piperouting capability might offer an attractive way to cut down
on dull and irksome work, leading to saving of time and
money. Pipe routing has traditionally been done largely by
eye (Wangdahl, Pollock, & Woodward, 1974), and developing pipe-routing programs has been constrained by various
considerations, such as:
(a)
(b)
(c)
(d)
(e)
obstacle avoidance
minimum equipment clearance requirements
accessibility of valves by hand or by reach-rods
maximization of support sharing with other pipelines
minimization of pipeline length and number of bends
2. Routing algorithms
2.1. Maze algorithm
Lee (1961) proposed a maze algorithm, also called Lees
algorithm or the grid expansion algorithm. This algorithm
posits a grid of cells. Obstacles are marked X on the cells
corresponding to them. The algorithm begins from a starting
cell S, and the cells adjoining that cell are labeled with a
0957-4174/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 9 5 7 - 4 1 7 4 ( 0 2 ) 0 0 0 4 9 - 0
300
J.-H. Park, R.L. Storch / Expert Systems with Applications 23 (2002) 299309
Table 1
Earlier studies on pipe routing
Algorithm
Network optimization
Au thor
Year
Dimension
Domain
Operation/Maintenance
Installation
Safety
Branch
Newell
1972
3D
General
X
X
X
P
Wangdahl
1974
2D
Shipbuilding
X
X
X
X
Maze
Guiradello
1993
3D
Plant
X
P
X
X
Zhu
1991
2D/3D
Robotics
X
X
X
X
Rourke
1975
3D
General
X
X
X
X
Mitsuta
1986
3D
General
X
X
F
X
Escape
GA
Schmidt
19
3D
Plant
X
P
X
X
Ito
1999
2D
General
X
P
X
X
3. Understanding constraints
J.-H. Park, R.L. Storch / Expert Systems with Applications 23 (2002) 299309
301
4. Cell-generation method
4.1. Background
Many early studies on routing have been carried out
either by means of cell-decomposition methods or by means
of optimization network techniques. A major problem of
cell-decomposition methods is the number of cells. As the
number of cells increases, the difficulty in dealing with them
grows exponentially. And the main difficulty of optimization network techniques is to define all nodes before an
302
J.-H. Park, R.L. Storch / Expert Systems with Applications 23 (2002) 299309
J.-H. Park, R.L. Storch / Expert Systems with Applications 23 (2002) 299309
303
304
J.-H. Park, R.L. Storch / Expert Systems with Applications 23 (2002) 299309
J.-H. Park, R.L. Storch / Expert Systems with Applications 23 (2002) 299309
305
styles can be built for the main pipeline Fig. 7(b). The
pipeline to the M/E Air Coolers has an end-forked form and
can be connected either to P001 or to P002 (Fig. 7(c)). Then,
the three-inch pipeline can be treated as a main pipeline, in
turn, and the two-inch pipeline is a middle-fork form (Fig.
7(d)).
high frequency bending is used for large pipes (Fig. 8). The
bigger the pipeline diameter, the more costly the bending,
because it takes more man-hours. As for installation cost, it
is directly related to pipe-support costs, for pipelines require
a series of pipe-supports. Pipe-support cost is comprised of
the costs for raw materials, paint, transportation, welding,
and u-bolts. The distance between supports varies in
accordance with the pipe size and the particularities of the
underlying steel structure.
5. Evaluation
First choice
Second choice
Third choice
2
3
4
6
8
10
12
14
16
18
20
24
0.25
0.38
0.50
0.75
1.00
1.25
1.50
1.75
2.00
2.25
2.50
3.00
0.78
1.17
1.56
2.34
3.12
3.90
4.68
5.46
6.24
7.02
7.80
9.36
1.89
2.84
3.79
5.68
7.57
9.46
11.36
13.25
15.14
17.03
18.93
22.71
O Co TotalHoursOfOperation HourlyWage
Operability coefficient Co . The operability coefficient Co
reflects the difficulties and discomfort inherent in operating
the valves. Difficulty is taken as proportional to handwheel
diameter. As for discomfort, the ABS guidelines (1998)
plotted in Fig. 9 show that the first choice is better than the
second choice and much better than the third choice.
Research by Karwowski and Salvendy (1998) found a linear
relationship between gravitational load and body discomfort
as shown in Fig. 10.
The numbers in the bars are the degree of comfort. The
degree of comfort for operating in the first choice zone can
be averaged as 2.65 and that for the second choice would be
0.65. In other words, the second choice is 3.12 times more
uncomfortable than the first choice, while the third choice is
7.57 times more uncomfortable. Table 2 shows the
operability coefficients. The underlying assumption is that
the operation of the valves for an eight-in pipe is taken as a
norm and assigned a value of 1. When valves are located
within the same zone with regard to a standing body, the
coefficient varies with handwheel diameter. Coefficients for
the second choice zone are derived by multiplying the first
choice coefficients by 3.12.
Total hours of operation. This factor is given as the
TimeRequiredToOperateEachTime times TotalNumOf
Operation. The term TotalNumOfOperation means the
total number of operations through a valves lifetime and
306
J.-H. Park, R.L. Storch / Expert Systems with Applications 23 (2002) 299309
than the N2 costs that the difference between the best case
of the N2 descendant nodes and the worst case of N1
descendant nodes could not be made up, the node N2 and
its dependents can be deleted. In Fig. 12, as B2 is low
enough, B1, B3 and their dependents are pruned.
Similarly, the algorithm deletes C3 and D2.
6. Implementation
6.1. Input data
The input data, mainly from system diagrams and
equipment arrangement drawings, has to do with boundaries, equipment, preferred operation area for valves, and
pipelines to be routed. Information on pipelines consists of
the pipeline number, diameter, directional specification, the
types and coordinates of endpoints, and branch pipeline
number. Daughter pipelines are given the mother pipeline
number, instead of an endpoint type.
J.-H. Park, R.L. Storch / Expert Systems with Applications 23 (2002) 299309
307
Description
2
10
14
20
21
22
23
24
30
35
40
55
56
65
66
68
308
J.-H. Park, R.L. Storch / Expert Systems with Applications 23 (2002) 299309
8. Conclusion
In this paper, a cell-generation driven pipe-routing
method has been developed to supply candidate paths.
From a tree of combinations, the algorithm chooses an
appropriate path for each pipeline from the candidate paths
after various non-geometric aspects, such as material costs,
installation costs, and valve operability, are assessed from a
fiscal point of view.
The methodology showing how various design constraints might be accommodated by suitable algorithms can
be a cornerstone for the evolution of design automation for a
domain fraught with complications. The algorithm presented here is for pipe routing. Further research will look
Acknowledgments
The authors would like to acknowledge David Evans,
Lawrence Pierce, and Randy Bird of Integration Partners,
Inc., San Diego, CA, for providing software and valuable
technical advice. They also would like to thank pipe-routing
experts, Jung-Hyun Park, Jin-Soo Youn, Young-Min Kang,
Pil-Joong Nam and Woo-Jong Kim of Samsung Heavy
Industries, Korea, for indispensable advice.
J.-H. Park, R.L. Storch / Expert Systems with Applications 23 (2002) 299309
References
The American Bureau of Shipping (ABS) (1998). Guidance notes on the
application of ergonomics to marine systems.
Guiradello, R. (1993). Optimization of process plant layout. PhD
Dissertation, University of Wisconsin: Madison.
Hightower, D. W. (1969). A solution to line routing problems on the
continuous plane. Proceedings of Sixth Design Automation Workshop,
IEEE (pp. 1 24).
ISO/CDC 10303-217 (1996). Product Data Representation and Exchange
Part: 217, Application Protocol: Ship Piping.
Ito, T. (1999). A genetic algorithm approach to piping route path planning.
Journal of Intelligent Manufacturing, 10, 103 114.
Kai-jian, S., & Hong-e, Z. (1987). Efficient routing algorithm. ComputerAided Design, 19(7), 375379.
Kang, S., Myung, S., & Han, S. (1999). A design expert system for autorouting of ship pipes. Journal of Ship Production, 15(1), 19.
Karwowski, W., & Salvendy, G. (1998). Ergonomics in manufacturing.
Dearborn, MI: Society of Manufacturing Engineers.
Kobayashi, Y., Wada, Y., & Kiguchi, T. (1986). Knowledge representation
and utilization for optimal route search. IEEE Transactions on Systems,
Man, and Cybernetics, SMC-16(3), 454 462.
Lee, C. Y. (1961). An algorithm for path connections and its applications.
IRE Transactions on Electronic Computers, EC-10, 346 365.
Mandl, C. (1979). Applied network optimization. New York: Academic
Press.
Mitsuta, T., Kobayashi, Y., Wada, Y., & Kiguchi, T. (1986). A knowledgebased approach to routing problems in industrial plant design.
Proceedings of the Sixth International Workshop Expert System and
Their Applications (pp. 237 253). France.
Newell, R. G. (1972). An interactive approach to pipe routing in process
plant. Information processing 71. Proceedings of IFIP Congress 71 (pp.
10801085). London.
Nicholson, T. A. J. (1966). Finding the shortest route between two points in
a network. The Computer Journal, 9(3), 275 280.
Radford, A. D., & Gero, J. S. (1988). Design by Optimization in
309