the effect of welding on the one-dimensional cutting-stock ...the effect of welding on the...
TRANSCRIPT
-
Research ArticleThe Effect of Welding on the One-DimensionalCutting-Stock Problem: The Case of Fixed FirefightingSystems in the Construction Industry
Kolos Csaba Γgoston 1,2
1Department of Operations Research and Actuarial Sciences, Corvinus University of Budapest, FoΜvaΜm teΜr 13-15,1093 Budapest, Hungary2Institute of Economics, Research Centre for Economic and Regional Studies, Hungarian Academy of Sciences,ToΜth KaΜlmaΜn u. 4, 1097 Budapest, Hungary
Correspondence should be addressed to Kolos Csaba AΜgoston; [email protected]
Received 5 October 2018; Revised 11 January 2019; Accepted 31 January 2019; Published 3 March 2019
Academic Editor: Viliam Makis
Copyright Β© 2019 KolosCsaba AΜgoston.This is an open access article distributed under theCreativeCommonsAttribution License,which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Safety is paramount in the construction industry and the fixed sprinkler and water spray systems used in firefighting involvenetworks of pipes of various lengths. Manufacturers of such fixed firefighting systems need to either cut the existing stocks tolengthβa (one-dimensional) cutting-stock problemβor lengthen the existing stocks or leftover segments through welding, a (one-dimensional) cutting-stock problem with welding. Best industry practice safety requirements allow only one weld per length ofpipe. The case of a Hungarian manufacturer of fixed firefighting systems motivates this article, which argues that the cutting-stock problem with welding (with single- or multiple-size stocks) may be converted to an equivalent cutting-stock problem (withmultiple-size stocks). Readily available algorithms and software may then be used to generate an optimal cutting plan for theequivalent cutting-stock problem. Subject to certain restrictions, the optimal cutting plan for the equivalent cutting-stock problemmay then be converted to cutting patterns for the original cutting-stock problem with welding.
1. Introduction
Safety is paramount in the construction industry and thefixed sprinkler and water spray systems used in firefightingare integral to a wide variety of buildings ranging fromwarehouses, through supermarkets, to educational establish-ments. Integrating fixed firefighting systems is in many casesmandatory, governed in the EU member countries by theEuropean Standard EN 12845, currently incorporated, forexample, in standards MSZ EN 12845:2015 in Hungary andBS EN 12845:2015 in the UK.
Fixed firefighting systems are assembled on site andinvolve networks of pipes (henceforth, pipes) of variouslengths (henceforth, sizes) manufactured from the pipes instock (henceforth, stocks). More often than not, manufactur-ers need to either cut the existing stocks to size (for example,from 6 meters to 4 meters from one stock, with a 2-meterleftover segment) or lengthen the existing stocks or leftover
segments through welding (for example, from 6 meters to 8meters from two stocks, with a 4-meter leftover segment).Leftover segments may be welded to other stocks or to eachother, but best industry practice safety requirements allowonly one weld per pipe.
In the case of aHungarianmanufacturer of fixed firefight-ing systems, the company uses single-size, 6-meter stocks toproduce pipes; Table 1 illustrates the sizes (14, in total) andquantities (31, in total) required for a particular network.Only one pipe size coincides with the stock size; 11 others areshorter and two are longer.
Manufacturing pipes for fixed firefighting systems isa one-dimensional cutting-stock problem with single- ormultiple-size stocks and welding restricted to one weld perpipe (henceforth, cutting-stock problem with welding). Thisarticle sets out to find its optimal solution (that is, optimalcutting-and-welding plan) and argues that the readily avail-able algorithms and software for the cutting-stock problem
HindawiAdvances in Operations ResearchVolume 2019, Article ID 6507054, 12 pageshttps://doi.org/10.1155/2019/6507054
http://orcid.org/0000-0002-6738-0592https://creativecommons.org/licenses/by/4.0/https://doi.org/10.1155/2019/6507054
-
2 Advances in Operations Research
Table 1: Illustrative pipe sizes and quantities for the case of aHungarian manufacturer of fixed firefighting systems.
Stock size (π1, in mm)6000
Pipe sizes (π π, in mm) Pipe quantities (ππ)2000 14000 14500 44660 24680 115000 15096 15250 15260 15500 15660 16000 27000 27200 2
with multiple-size stocks (henceforth, cutting-stock problem)may be used for the cutting-stock problem with welding.Numerical testing found the algorithm suggested in thisarticle (see Section 5) adequate for the case of a Hungarianmanufacturer of fixed firefighting systems.
Following this brief introduction, Section 2 reviews therelevant literature and positions this article in the ongo-ing debate fuelling research in the cutting-stock problem.Section 3 formulates the cutting-stock problem and thecutting-stock problem with welding as a mixed integer linearprogram, discusses cutting-and-welding pattern properties,and proves that optimal cutting-and-welding plans consistof minimal cutting-and-welding patterns. This characteristicof optimal cutting-and-welding plans is fundamental to thealgorithm suggested in the next section. Section 4 discussesthe practical difficulties caused by high cutting-and-weldingstock quantities and nonsequential cutting-and-welding pat-terns. The ensuing need to limit the number of cutting-and-welding stocks and thus restrict the cutting-stock problemwith welding shapes the algorithm suggested in this sec-tion. Generating minimal cutting-and-welding patterns andoptimal cutting-and-welding plans pose possible difficulties,also explored in this section, alongside program runningtimes, sequential patterns, and decision maker preferences.In Section 5 we apply the algorithm suggested in the previoussection to the case of a Hungarian manufacturer of fixed fire-fighting systems with satisfactory results. Finally, Section 6summarises the findings and suggests directions for furtherresearch.
2. Background
The cutting-stock problemβthe question of how to opti-mise stock or how to cut stock to minimise costβwasfirst investigated by Kantorovich [1] and Kantorovich [2].
The column-generation algorithm suggested by Gilmore andGomory [3] and Gilmore and Gomory [4] more than twentyyears later became widely used by researchers and practition-ers alike; Gilmore and Gomory [3] and Gilmore and Gomory[4] βwere the first to present techniques practically applied todifficult real-world problemsβ (Sweeney and Paternoster [5]).Complex cutting-stock problems generate complex cuttingplans with very large numbers of feasible cutting patterns.Instead of attempting to generate and investigate them allat once, the Gillmore and Gomory algorithm first generatesand investigates only a few. Subsequently generated cuttingpatterns replace existing ones if proven more efficient uponinvestigation, and through the iterative process the algorithmarrives at an optimal solution.
The strength of the Gillmore and Gomory algorithm liesin its ability to generate an optimal solution; its weaknessin its inability to generate an optimal cutting plan wherecutting pattern weights are integers. More often than not,rounding decimal cutting pattern weights to the nearestinteger results in infeasible or suboptimal cutting plansand/or overproducing some orders and underproducingothers. Generating an optimal cutting plan for the cutting-stock problem thus became the focus of further research;Roodman [6], WaΜscher and Gau [7], and Gradisar andTrkman [8] suggested heuristic methods, while Dyckhoff [9],Scheithauer and Terno [10], Vance et al. [11], Vanderbeck [12],and Belov and Scheithauer [13] proposed further algorithms.Sweeney and Paternoster [5] andWaΜscher et al. [14] providedcomprehensive literature reviews of these and otherβusuallyheuristicβmethodologies.
The cutting-stock problem therefore stayed in operationsresearch focus even in its simplest, original, one-dimensionaldefinition; in recent years, for example, Levine and Ducatelle[15] and ArauΜjo et al. [16] used evolutionary algorithm forthe cutting-stock problem. Particularly relevant to this articlethough are the works of Cui and Yang [17], who investigatedleftover segment recycling.
Cutting stock is a nondeterministic polynomial-time-(NP-) hard problem (Cheng et al. [18]), which may requireexponential software running times; nonetheless, in practice,even large samples may be investigated for optimal solutions(see Goulimis [19]; Belov and Scheithauer [13]). Thereforethis article assumes that there are algorithms and softwarefor obtaining the exact (global) solution of a cutting-stockproblem as defined in Section 1 (see WaΜscher et al. [14] fortypology, includingMSSCSP). If the goal is to design businessapplications where suboptimal solutions are acceptable, thenexact (global) solutions are not necessary, because good(heuristic) solutions are sufficient.
Zak [20] defines the skiving stock problem, a counterpartto the cutting-stock problem, as βfind the maximum numberof items with minimum length πΏ that can be constructed byconnecting a given supply of smaller item lengthsβ (Marti-novic and Scheithauer [21]).
In the skiving stock problem, the number of connecteditems is not limited. In our case, it is crucial that every itemcan be constructed from two pieces at most. The investigatedcutting-and-welding problem is a combination of the cutting-stock problem and the skiving stock problem. Tanir et al.
-
Advances in Operations Research 3
Pattern A; (j, u1, u2, u3)=(1, 1, 0, 1)Pattern B; (j, u1, u2, u3)=(1, 3, 0, 0)Pattern C; (j, u1, u2, u3)=(1, 1, 1, 0)Pattern C ; (j, u1, u2, u3)=(1, 1, 1, 0)Pattern D; (j, u1, u2, u3)=(1, 1, 0, 0)
Figure 1: Some feasible patterns for cutting 2-, 3-, and 4-meter pipes from single-size, 6-meter stocks.
[22] examined the possibility of welding in a cutting-stockproblem and suggested a heuristic method. There were norestrictions on the number of welds per pipe, while in thispaper the number of welds per pipe is restricted to one.
3. Theory
3.1. Formulating the Cutting-Stock Problem. The cutting-stock problem consists of generating the optimalβminimalcostβcutting plan. πΌ pipes of various sizes π π (π = 1, 2, . . . πΌ)need to be cut in various quantities ππ from π½ stocks of varioussizes ππ (π1, π2, . . . ππ½). Feasible cutting patterns may berepresented by sets of integers (π, π’1, π’2, . . . π’πΌ), where π is thestock index and π’π is the number of pipes of size π π producedfrom stock ππ. Evidently, βπΌπ=1 π’ππ π β€ ππ and βπΌπ=1 π’π β₯ 1.
For example, π 1 = 2-, π 2 = 3-, and π 3 = 4-meter pipescould be cut from single-size, π1 = 6-meter stocks in variousfeasible patterns (π, π’1, π’2, π’3). Figure 1 illustrates five suchcutting patterns: the 6-meter stocks are cut into one 2-meterpipe and one 4-meter pipe (6 = 2 + 4), in pattern π΄ (1, 1, 0,1); into three 2-meter pipes (6 = 2 + 2 + 2), in pattern π΅ (1,3, 0, 0); into one 2-meter pipe, one 3-meter pipe, and one 1-meter leftover segment (6 = 2 + 3 + 1), in pattern πΆ (1, 1, 1, 0);into one 2-meter pipe, one 3-meter pipe, and two 0.5-meterleftover segments (6 = 0.5+2+3+0.5), in patternπΆ (1, 1, 1, 0);and into one 2-meter pipe and one 4-meter leftover segment(6 = 2 + 4), in pattern π· (1, 1, 0, 0). Cutting patterns reflectwhat is produced, not how is produced; as such, patterns πΆand πΆ are equivalent and patternπ· is wasteful.
If P represents all the feasible cutting patterns, thencutting pattern πβ β P produces π’πβ pipes of size π π at thepattern price πβ, the stock price, that is, if the cutting cost isignored as insignificantly small.
Thus the cutting-stock problemmay be formulated as thefollowing mixed integer linear program:
minbββ:πββP
πβπβs.t. β
β:πββP
πβπ’πβ β₯ ππ βππβ βN,
(1)
where every feasible solution represents a cutting plan, anoptimal cutting plan is a feasible solution with minimal cost,and the decision variable πβ βN indicates how many times acutting plan uses cutting pattern πβ.
Due to its exponential size, solving the mixed integerlinear program is not practical for large instances. In practice,this should be donewith a combination of branch-and-bound
and column-generation (see Belov and Scheithauer [13] forbranch-and-price and column-generation).
Optimal cutting plans do not exclude wasteful cuttingpatterns like pattern π· in Figure 1; and a cutting-stockproblem may have more than one optimal cutting plan. Forexample, if four 2-meter pipes need to be cut from single-size,6-meter stocks, then one optimal cutting plan could use twopatterns to cut one 6-meter stock in three 2-meter pipes (6 =3 Γ 2) and another 6-meter stock in one 2-meter pipe andone4-meter leftover segment (6 = 2 + 4). Another optimal plancould use the first pattern twice to cut two 6-meter stocks insix 2-meter pipes (2 Γ 6 = 6 Γ 2), two 2-meter pipes toomany. Subject to additional considerations such as numberof cuts, amount of waste and size of overproduction, decisionmakers may prefer one optimal cutting plan to another.
3.2. Formulating the Cutting-Stock Problem with Welding.Similarly to the cutting-stock problem, the cutting-stock problem with welding consists of generating theoptimalβminimal costβcutting-and-welding plan. Thenumber of feasible cutting-and-welding patterns is just largerthan the number of feasible cutting patterns. For example,if 4-meter pipes need to be produced from single-size,6-meter stocks, then only one cutting pattern is feasiblein the cutting-stock problem as opposed to two cutting-and-welding patterns feasible in the cutting-stock problemwith welding. In the cutting-stock problem, the 6- meterstocks can only be cut into 4-meter pipes and wasteful2-meter leftover segments. In the cutting-stock problem withwelding, the βwastefulβ 2-meter leftover segments may bewelded together, in additional 4-meter pipes (see patterns π΄and π΅ in Figure 2). However, whether pattern π΅ is preferableto pattern π΄ depends on the price per weld relative to theprice per stock, if higher, then pattern π΄ prevails.
In the cutting-stock problem with welding, feasiblecutting-and-welding patterns may be represented by sets ofintegers (U, u) = (π1, π2, . . . , ππ½, π’1, π’2, . . . , π’πΌ) and uniqueassignment matrices π΄ with βπ½π=1ππ rows, βπΌπ=1 π’π columns,and only zero and one as elements.ππ represents the numberof stocks of size ππ, π = 1, . . . , π½, used to produce π’π pipes ofsize π π, π = 1, . . . , πΌ, through cutting and welding. Thus, thefirst row represents the first stock of size π1, the second rowthe second stock of size π1,. . ., and theπ1th row theπ1th stockof size π1. Similarly, the (π1+1)th row represents the first stockof size π2, the (π1 + 2)th row the second stock of size π2,. . .,the (π1 +π2)th row theπ2th stock of size π2, and so on. Also,the first column represents the first pipe of size π 1, the secondcolumn the second pipe of size π 1,. . ., and the π’1th column theπ’1th pipe of size π 1. Similarly, the (π’1+1)th column representsthe first pipe of size π 2, the (π’1+2)th column the secondpipe of
-
4 Advances in Operations Research
Pattern A; (U1, u1)=(1, 1), A=(1) Pattern B; (U1, u1)=(2, 3) A = (1 1 00 1 1)
Figure 2: Patterns for producing 4-meter pipes from single-size, 6-meter stocks through cutting and cutting and welding.
size π 2,. . ., the (π’1 +π’2)th column the π’2th pipe of size π 2, andso on. If rows are indexed by π, columns by π, and elementsby πππ, then πππ = 1 implies that at least some segment ofpipe π is produced with at least some segment of stock π. Thisarticle assumes that no stock size and no stock is redundantand that therefore πππ = 1 at least once in every row and oncein every column. Since any one stock may be assigned to anynumber of pipes, but any one pipemay only be produced fromtwo stocks at the most, πππ = 1 no more than twice in everycolumn.
The set of integers (U, u) and assignment matrix π΄represent a feasible cutting-and-welding pattern if an equal-size weldingmatrixπ can be found to offer a feasible weldingoption. If rows are indexed by π, columns by π, and elementsby π€ππ, then π€ππ > 0 if and only if πππ = 1. Also, if π(π) is thesize of the stock represented by row π and π (π) is the size of thepipe represented by column π, then the sumof all the elementsin any one row is less than π(π) and the sumof all the elementsin any one column is equal to π (π). Furthermore,βπ½π=1ππ β₯ 1and βπΌπ=1 π’π β₯ 1. In addition, one has the following:
(i) A pipe size shorter than the sumof all the stock sizes isinsufficient in itself. For example, if π1 = 1meter, π2 =6meters, and π 1 = 4meters, then (π1, π2, π’1)=(4, 0, 1)requires three welds and is not a valid cutting-and-welding pattern.
(ii) Cutting-and-welding patterns reflect what is pro-duced, not how is produced. For example, from acutting-stock problem with welding point of view,patterns πΆ and πΆ in Figure 1 have identical setsof integers (π1, π’1, π’2, π’3)=(1,1,0,1) and assignmentmatrices π΄ = (1 1) and welding matrices π =(2000 3000)) (see Section 3.1).For another example, 3.6-meter pipes can be pro-duced from single-size, 6-meter stocks in variouscutting-and-welding patterns, for example, by cuttingone stock in one 3.6-meter pipe and one 2.4-metersegment (6 = 3.6 + 2.4), by cutting another stockin one 3.6-meter pipe and two 1.2-meter segments(6 = 1.2+3.6+1.2), by welding the 2.4-meter segmentwith one of the two 1.2-meter segments (2.4 + 1.2 =3.6), with (π1, π’1)=(2, 3), π΄ = ( 1 1 00 1 1 ), and π =( 3600 2400 00 1200 3600 ), by cutting each of two stocks in one3.6-meter pipe, one 1.8-meter segment, and one 0.6-meter leftover segment (2 Γ 6 = 2 Γ 3.6+2 Γ 1.8+2 Γ 0.6), and by welding the two 1.8-meter segments(1.8+1.8 = 3.6), with identical (π1, π’1)=(2, 3) andπ΄ =( 1 1 00 1 1 ) as before, but different π = ( 3600 1800 00 1800 3600 ),and so on.
(iii) Redundancies can appear in cutting-and-weldingpatterns, but these do not affect optimal solutions.
For example, by cutting each of two 6-meter stocks inone 3.6-meter pipe, one 1.8-meter segment, and one0.6-meter leftover segment (2 Γ 6 = 2 Γ 3.6 + 2 Γ1.8+2 Γ 0.6), one in three 3.6-meter pipes is welded.Six identical sets of integersβ(π1, π’π)=(2, 3)βandsix different assignment matricesβπ΄1 = ( 1 1 00 1 1 ),π΄2 = ( 1 1 01 0 1 ), π΄3 = ( 1 0 11 1 0 ), π΄4 = ( 0 1 11 1 0 ), π΄5 =( 0 1 11 0 1 ), andπ΄6 = ( 1 0 10 1 1 )βcorrespond to six differentcutting-and-welding patterns with identical practicaloutcomes.
(iv) Welding is not a compulsory component of cutting-and-welding patterns. For example, two 5-meter pipescan be produced from two 6-meter stocks by cuttingeach of the two 6-meter stocks in one 5-meter pipeand one 1-meter leftover segment (2 Γ 6 = 2 Γ 5 +2 Γ 1), with (π1, π’1)=(2, 2) and π΄ = ( 1 00 1 ) as if thecutting patterns (π1, π’1)=(1, 1) and π΄=(1) were usedtwice.
(v) The number of cutting-and-welding patterns is finite,since the number of pipes and the number of stocksare both finite.
(vi) Any one pipe can only be produced from two stocksat the most, but a cutting-and-welding pattern mayinvolve more than two stocks, which may be cut andwelded in various ways.
For example, six 5-meter pipes can be produced from five6-meter stocks by cutting each of two 6-meter stocks in one 5-meter pipe and one 1-meter segment (2 Γ 6 = 2 Γ 5+2 Γ 1),by cutting each of two 6-meter stocks in one 4-meter segmentand one 2-meter segment (2 Γ 6 = 2 Γ 4+2 Γ 2), by cuttingone 6-meter stock in two 3-meter segments (6 = 2 Γ 3),by welding each of the two 1-meter segments to one of thetwo 4-meter segments (2 Γ 1 + 2 Γ 4 = 2 Γ 5), and bywelding each of the two 2-meter segments to one of the two3-meter segments (2 Γ 2 + 2 Γ 3 = 2 Γ 5) (see Figure 3).This cutting-and-welding pattern is therefore represented by(π1, π’1)=(5, 6),
π΄ =(((
1 1 0 0 0 00 1 1 0 0 00 0 1 1 0 00 0 0 1 1 00 0 0 0 1 1))),
-
Advances in Operations Research 5
π =(((
5000 1000 0 0 0 00 4000 2000 0 0 00 0 3000 3000 0 00 0 0 2000 4000 00 0 0 0 1000 5000)))
(2)
If Q represents all the feasible cutting-and-welding pat-terns, then cutting-and-welding pattern πβ produces π’πβ pipesof size π π at the pattern price πβ, the stock and welding prices,that is, if the cutting cost is ignored as insignificantly small.
Similar to the cutting-stock problem, the cutting-stockproblem with welding may be formulated as the followingmixed integer linear program:
minbββ:πββQ
πβπβs.t. β
β:πββQ
πβπ’πβ β₯ ππ βππβ βN,
(3)
where every feasible solution represents a cutting-and-welding plan, an optimal cutting-and-welding plan is afeasible solution with minimal cost, and the decision variableπβ βN indicates howmany times a cutting-and-welding planuses cutting-and-welding pattern πβ.
Similar to the cutting-stock problem, due to its expo-nential size, this method is not practical for solving largeinstances. In practice, this should be replaced with a com-bination of branch-and-bound and column-generation tech-niques (see Belov and Scheithauer [13], for example, forbranch-and-price and column-generation techniques).
3.3. The Relation between Optimal Cutting-and-Welding Plansand Minimal Cutting-and-Welding Patterns
Proposition 1. Every optimal cutting-and-welding plan canbe constructed using cutting-and-welding patterns where thenumber of welds involved is one fewer than the number of stocksused. We call these cutting-and-welding patterns minimal.
Proof. Let us prove this proposition by contradiction andassume that there exists an optimal cutting-and-weldingplan with a cutting-and-welding pattern with π stocks (rowsπ1, π2, . . . , ππ in the assignment matrix π΄) and π€ welds, whereπ€ β₯ π. This cutting-and-welding pattern may be representedas a graphβa network of nodes and linking edges, wherethe nodes represent the stocks and the edges represent theweldsβwhere theremay be no edge, one edge, or two ormoreedges between any two nodes. If π€ β₯ π, then the networkis cyclical; let π1, π2, . . . , ππ be the stocks that form the cycleπ1 β π2 β β β β β ππ β π1. Corresponding indices(columns π1, π2, . . . , ππ in the assignment matrix π΄) need tobe found so that all the cyclical matrix elements ππ1π1 , ππ1π2 ,ππ2π2 , ππ2π3 , ππ3π3 , ππ3π4 . . ., πππβ1ππ , πππππ , and ππππ1 are one. If π
Figure 3: Patterns for cutting-and-welding six 5-meter pipes fromfive 6-meter stocks.
is the smallest among π€π1π1 , π€π2π2 ,. . ., π€ππππ elements in thecorresponding welding matrixπ, then subtract π from all ofthese elements and adding π to all the elementsπ€π1π2 ,π€π2π3 ,. . .,π€ππβ1ππ , and π€πππ1 results, among others, in a zero element,in other words, a pipe without welding and a cutting-and-welding pattern with one fewer welds. Modifications in πlead of course to modifications in π΄; if π€ππ > 0, then πππ = 1;if π€ππ = 0, then πππ = 0; and if π€ β 1 β₯ π, then the processrequires as many iterations as necessary to obtain a cycle-free graph, a tree or a forest. A contradiction is that since atree where the number of welds involved is one fewer thanthe number of stocks used represents a minimal cutting-and-welding pattern.
This characteristic of optimal cutting-and-welding plansplays a fundamental role in the algorithm suggested in thisarticle (see Section 4.1).
Important to note that Proposition 1 does not state thatwelds are always necessary. If we avoid welds, the number ofused stock is 1, and the number of welds is 0, so Proposition 1holds.
4. Calculation
By adding their lengths, π single- or multiple-size cutting-and-welding stocks may be converted to one equivalentcutting-stock and the cutting-stock problem with weldingmay be converted to an equivalent cutting-stock problem.If ππ€ is the price per weld, then the price of the equivalentcutting stock is the price of the π cutting-and-welding stocksplus (π β 1)ππ€. Multiple-size cutting-and-welding stocks (ofπ½ sizes) may generate large numbers of equivalent cuttingstocks, given by the combination ( π+π½β1
π). If β+β represents
one stock and β/β the border between two stock sizes, then,for example, β++/+//+++β represents two stocks from the firstsize, one from the second, none from the third, and threefrom the fourth. For π cutting-and-welding stocks and π½cutting-and-welding stock sizes, there are π½ β 1 β/β signs andπ+ π½β1 places where the π β+β signs could go. In the exampleabove, π = 6, π½ = 4, and ( 96 ) = 84.
Cutting-and-welding patterns are not necessarily sequen-tial; for example, the first stock could be cut in two segments,one to be welded with a segment from the fourth stock andthe other with a segment from the tenth stock. High cutting-and-welding stock quantities and nonsequential cutting-and-welding patterns cause practical difficulties that may be
-
6 Advances in Operations Research
reduced by reducing the cutting-and-welding stock quanti-ties.
Let us therefore limit π to π , and thus bound the cutting-stock problem with welding; the smaller the π , the smallerthe number of equivalent cutting patterns. For single-sizecutting-and-welding stocks (as in the case which motivatesthis article), the equivalent cutting-stock problem toleratesquite large sizes of π (see Section 4.1). Generally, however,finding the optimal, smallest π is of obvious importance (seeSection 4.3).
4.1. An Algorithm for Solving the Bounded Cutting-StockProblem with Welding. To solve the bounded cutting-stockproblem with welding, this article suggests a three-stepalgorithm thatβat the leastβis a good heuristic and thatβatthe bestβgenerates an optimal cutting-and-welding plan:
(1) Convert the bounded cutting-stock problem withwelding to an equivalent cutting-stock problem asfollows:
(a) Choose π β€ π single- or multiple-size boundedcutting-and-welding stocks in all the ways pos-sible; for example, 5- and 6-meter cutting-and-welding stocks and π = 3 generate two boundedcutting-and-welding stocks for π = 1 (5 and 6meters); three for π = 2 (5 + 5 = 10, 5 + 6 =11, and 6 + 6 = 12 meters); and four for π = 3(5 + 5 + 5 = 15, 5 + 5 + 6 = 16, 5 + 6 + 6 = 17, and6 + 6 + 6 = 18 meters).
(b) Add their lengths, to convert the boundedcutting-and-welding stocks to one equivalentcutting stock.
(c) Add their prices to the (π β 1)ππ€ welding price,to obtain the equivalent cutting-stock price.
(2) Solve the equivalent cutting-stock problem (that is,generate the optimal equivalent cutting plan) withthe aid of open-source or commercial software (seeSection 3.1).
(3) Convert each equivalent cutting pattern to a boundedcutting-and-welding pattern (henceforth, pattern con-version problem, see Section 4.2).
This algorithm poses two possible difficulties.First (see step (1)), multiple-size bounded cutting-and-
welding stocks may generate alternative equivalent cuttingstocks. For example, if the multiple-size bounded cutting-and-welding stocks consist of 4- and 6-meter stocks, thenthree 4-meter stocks (3 Γ 4 = 12) with two welds and two6-meter stocks (2 Γ 6 = 12) with one weld generate twoalternative equivalent cutting stocks of identical, 12-metersizes, but nonidentical prices. To solve the equivalent cutting-stock problem (see step (2)), good practice suggests that thealgorithm should consider the cheapest equivalent cuttingstock.
Second (see step (3)), to generate an (optimal) boundedcutting-and-welding plan, the optimal equivalent cuttingplan would need to be converted to minimal bounded
cutting-and-welding patterns (see Section 4.2).This difficultyis trivial in most cases, but by no means in all. For example,if 10-meter pipes need to be produced from single-size, 6-meter cutting-and-welding stocks, then five 6-meter stocksand four welds would generate a 30-meter equivalent cuttingstock and two cuts would generate three 10-meter pipes (5 Γ6 = 30 = 3 Γ 10). However, since one of the pipes wouldhave two welds instead of the maximum one, this cutting-and-welding pattern is not even feasible let alone minimal.The algorithm would not only need to solve the equivalentcutting-stock problem (see Section 3.1); it would also need totackle extreme cases, with nominimal bounded cutting-and-welding patterns (see Section 4.2).
4.2. Formulating the Pattern Conversion Problem. Let usconsider one of the equivalent cutting patterns with πΎsingle- or multiple-size stocks and πΏ single- or multiple-sizepipesβπ = 1, 2, . . . , πΎ and β = 1, 2, . . . , πΏβgenerated at step(2) (see Section 4.1). Also, let ππ be the size of stock π, π₯βπ thedecision variable which shows the size of segment of pipe βproduced from stock π, and π΅βπ β {0; 1} the binary variablewhich shows whether π₯βπ is zero or positive.
Thus the pattern conversion problem may be formulatedas the following mixed integer linear program:
πΏββ=1
πΎβπ=1
π΅βπ β mins.t.: π₯βπ β πππ΅βπ β€ 0 β = 1 . . . πΏ, π = 1 . . . πΎ
πΎβπ=1
π₯βπ = π β β = 1 . . . πΏπΏββ=1
π₯βπ β€ ππ π = 1 . . . πΎπΎβπ=1
π΅βπ β€ 2 β = 1 . . . πΏπ₯βπ β₯ 0 β = 1 . . . πΏ, π = 1 . . . πΎπ΅βπ β {0; 1} β = 1 . . . πΏ, π = 1 . . . πΎ
(4)
Converting equivalent cutting patterns to minimalbounded cutting-and-welding patterns may seem trivial andin most cases indeed it is. However, multiple-size stocks andlong pipes which all need to be welded complicate conversionand an equivalent cutting pattern might not convert to aminimal bounded cutting-and-welding pattern.
The value of the objective function indicates the numberof stocks and/or stock segments necessary to produce the πΏpipes. Each pipe contains at least one stock or stock segment,but not more than two; thus, the value of the objectivefunction is given by the number of pipes and the number ofwelds:
(i) Thus if the value is πΏ + πΎ β 1, then the boundedcutting-and-welding pattern generated is minimal(see Section 3.3).
-
Advances in Operations Research 7
(ii) If the value is less than πΏ + πΎ β 1, then the softwareused at step (2) (see Section 4.1) was inadequate,the equivalent cutting plan was suboptimal, and thebounded cutting-and-welding pattern generated isfeasible, but not minimal. In practice, albeit withno firm claim to optimality, this feasible patternmay be broken down into smaller, minimal pat-terns; after all, every pipe can be produced fromonly two stocks at the most; tree graphs such asthose discussed in Section 3.3 are noncyclical evenfor number of welds smaller than πΎ β 1. Sincethe number of stocks is identical, but the numberof welds is smaller, this set of minimal patternsis cheaper than the feasible pattern generated bythe pattern conversion problem. However, furthertheoretical investigation is outside the scope of thisarticle.
(iii) If the pattern conversion problem does not havea feasible solution, then the software used at step(2) (see Section 4.1) is not necessarily inadequate.In practice, albeit with no firm claim to optimal-ity, a feasible bounded cutting-and-welding patternmay be obtained through iteration by increasingthe number of stocks in the pattern conversionproblem from πΎ to πΎ + 1 and πΎ + 2, and so on.In the case of multiple-size stocks, increasing thenumber of longest-size stocks will definitely resultin a feasible solution; in certain cases, increasingthe number of shorter-size stocks may result in afeasible solution too. As before, this feasible pat-tern with πΎ stocks and fewer than πΎ β 1 weldsmay be broken down into smaller, minimal pat-terns.
4.3. The Relation between Optimal Equivalent Cutting Plans,Optimal Bounded Cutting-and-Welding Plans, and Opti-mal Cutting-and-Welding Plans
Proposition 2. If every equivalent cutting pattern of theoptimal equivalent cutting plan can be converted to a mini-mal bounded cutting-and-welding pattern, then the algorithmgenerates an optimal solution for the bounded cutting-stockproblem with welding.
Proof. Let us prove this proposition by contradiction andassume that the algorithm does not generate an opti-mal solution for the bounded cutting-stock problem withwelding, although every equivalent cutting pattern of theoptimal equivalent cutting plan can be converted to aminimal bounded cutting-and-welding pattern. Then thereexists a bounded cutting-and-welding plan cheaper than thebounded cutting-and-welding plan generated by the algo-rithm. Since all the bounded cutting-and-welding patternsof this bounded cutting-and-welding plan can be convertedto equivalent cutting patterns, then the equivalent cuttingplan generated is cheaper than the optimal equivalent cut-ting plan generated by the algorithm; that is a contradic-tion.
Proposition 3. The optimal solution for the bounded cutting-stock problem with welding is an optimal solution for thecutting-stock problem with welding too if π is sufficiently large.Proof. The algorithm suggested in this article for solvingcutting-stock problems with welding relies on limiting theavailable cutting-and-welding stocks to π by bounding thecutting-stock problem with welding (see Section 4.1). Sinceevery pipe can be produced from two stocks at the most, π has an upper limit of 2βπΌπ=1 ππ.
However, this upper bound may still be too large forpractical purposes and finding the lowest upper bound isimportant (the lower the upper bound, the fewer the numberof equivalent cutting stocks), but problematic even for single-size cutting-and-welding stocks. Moreover, even the lowestupper bound may be too large for practical purposes. Forexample, let us consider producing π1 1.001-meter pipes fromsingle-size, 1-meter cutting-and-welding stocks. If π1 =1,000,then π β₯ 2βπΌπ=1 ππ = 2 Γ π1 = 2Γ 1,000 = 2,000, butthe lowest upper bound is evidently 1,001, if the cuttingcost is ignored as insignificantly small. Finding the lowestupper bound is therefore possible, in special cases, butgenerally problematic, particularly for multiple-size cutting-and-welding stocks.
4.4. Algorithm Running Times. The algorithm running timeconsists of the running time for the equivalent cutting-stockproblem (see step (2), Section 4.1) and the running time forthe pattern conversion problem (see step (3), Section 4.1).
To investigate running times for the equivalent cutting-stock problem, the BarCut (version 2.1) program was run ona computer with an Intel Duo 2.33GHz processor, 6GB ofrandom access memory (RAM), and Windows 7 Enterpriseas operating system. At less than 9 seconds, the running timesfor the case which motivates this article proved acceptable(see Section 5).
To investigate running times for the pattern conversionproblem, the GLPK (version 4.55) program was run onthe same computer (see Table 2, where the second columnshows the equivalent cutting patternsβgenerated by a com-plex, hypothetical solution of the equivalent cutting-stockproblemβthat need to be converted to feasible cutting-and-welding patterns). If the pattern conversion problem did nothave an optimal solution, the running times proved short; ifit did, the running times proved very long at times and inobvious need of further analysis.
Moderate numbers of pipes (see the second columnin Table 2) and moderate numbers of single-size, 6-metercutting-and-welding stocks (see the third column in Table 2)generate feasible solutions in (milli)seconds (see columnG1 in Table 2). Large numbers may require long runningtimes, but feasible solutions can generally be generated in(milli)seconds and the actual value of the objective function(βπΏβ=1βπΎπ=1 π΅βπ) indicates the existence of a feasible solution(see Section 4.2). The βnormalβ, optimal value of the objectivefunction is πΏ+πΎβ1 (see Section 4.2); however, more efficientcutting-and-welding patterns may be generated, with fewer
-
8 Advances in Operations Research
Table 2: Algorithm running times for minimal cutting-and-welding patterns (β = no minimal solution).Pattern Pipes (π β, in mm) Stocks G1 G2 G3 G4 G5 S1(πΎ)A 2070, 2140, 2210, 2280, 2350, 2420, 4530 3 0.4 0.1 0.1 0.1 0.1 0.1
B 2070, 2140, 2210, 2280, 2350, 2420, 2490,2560, 2630, 2850 4 > 3600 0.1 0.1 0.1 0.1 0.1C 2070, 2140, 2210, 2280, 2350, 2420, 2490,2560, 2630, 2700, 2770, 3380 5 > 3600 0.1 0.1 0.1 0.1 0.1D 2070, 2140, 2210, 2280, 2350, 2420, 2490,2560, 2630, 2700, 2770, 2840, 2910, 3630 6 > 3600 0.1 0.1 0.1 > 3600 0.1E
2070, 2140, 2210, 2280, 2350, 2420, 2490,2560, 2630, 2700, 2770, 2840, 2910, 2980,
3050, 36007 > 3600 0.1 0.1 0.1 > 3600 0.2
F2070, 2140, 2210, 2280, 2350, 2420, 2490,2560, 2630, 2700, 2770, 2840, 2910, 2980,
3050, 3120, 3190, 32908 > 3600 0.4 0.6 1.7 > 3600 0.2
G2070, 2140, 2210, 2280, 2350, 2420, 2490,2560, 2630, 2700, 2770, 2840, 2910, 2980,
3050, 3120, 3190, 32608 > 3600 - - - - -
H 10000, 10000, 10000 5 0.1β - - - - -I 11000, 11000, 11000, 3000, 3000, 3000 7 86.6 0.0 0.0 1.0β - 0.1βJ
1000, 1000, 1000, 1000, 1000, 1000, 1000,1000, 1000, 11001, 11001, 11001, 11001, 11001,
11001, 11001, 11001, 11493, 1149920 > 3600 118.4 > 3600 > 3600 > 3600 > 3600
thanπΎβ1welds (see patternG in Table 2, withπΎβ2 = 8β2 =6 welds). (Pattern H in Table 2 has no minimal solution.)Even withβπΏβ=1βπΎπ=1 π΅βπ = πΏ+πΎβ1 as constraints, where
appropriate (see Section 4.4.1), the algorithm running timefor the optimal solution is generally amatter of (milli)seconds(see column G2 in Table 2).
4.4.1. DecisionMaker Preferences and Sequential Cutting-and-Welding Patterns. Equivalent cutting patterns may convertto alternative cutting-and-welding patterns and decisionmakers may prefer some to others. Preferences may beeasily incorporated by changing the objective function fromβπΏβ=1βπΎπ=1 π΅βπ (see Section 4.2) to βπΏβ=1 πβ(βπΎπ=1 π΅βπ), whereπβ represents decision makerβs score point (that is, preferencelevel) for pipe β; lower πβ values indicate higher preferencesfor welding. Since longer pipes are likely to be welded fromstocks, not just cut, we may logically attribute an π1 = 0to the longest pipe, an π2 = 1 to the second-longest pipe,. . ., and an ππΏ = πΏ β 1 to the shortest pipe. Without theβπΏβ=1βπΎπ=1 π΅βπ = πΏ + πΎ β 1 constraints, the program runningtimes for the pattern conversion problem might be longand an alternative, two-step method preferable. In step one,the program is allowed to run for the few (milli) secondsnecessary to calculate the minimal number of welds; arrivingat the optimal value of the objective function is a fast process;investigating all the possible alternatives is not.
In step two, the objective function is changed fromβπΏβ=1βπΎπ=1 π΅βπ to βπΏβ=1 πβ(βπΎπ=1 π΅βπ) and the βπΏβ=1βπΎπ=1 π΅βπ =πΏ + πΎ β 1 constraints or alternative constraints, based onstep one, are added to generate a preferred pattern hopefullyquickly (see column G3 in Table 2). Pattern J in Table 2
represents an exceptionally difficult case, where a minimalpattern can be generated relatively quickly only if the decisionmakerβs preferences are disregarded.
Understandably, decision makers prefer sequentialcutting-and-welding patterns, where pipes are placed inrelevant order and the end of one stock is welded to thebeginning of the next. In the tree graph discussed inSection 3.3, sequential patterns would correspond to paths.
When all the pipes are shorter than the shortest stock,sequential patterns are evidently possible; in fact, everypermutation yields a sequential pattern. When some of thepipes are longer than the shortest stock, sequential patternsare not always possible. For example, if 4-, 5-, and 9-meterpipes need to be produced from single-size, 6-meter stocks,then the permutation (9, 5, 4) generates a sequential patternwhile the permutation (5, 9, 4) does not, because the 9-meterpipe would require two welds. For another example, thereis no sequential pattern if three 11-meter and three 3-meterpipes need to be produced from single-size, 6-meter stocks,but there is a minimal, nonsequential pattern involving sevenstocks, six welds, and five cuts (see Figure 4). (Six stocks, threewelds, and three cuts will generate three 11-meter pipes andthree 1-meter segments.The seventh stock will be cut in three2-meter segments, each to be welded to one of the three 1-meter segments to generate three 3-meter pipes.)
Sequential cutting-and-welding patterns may be gener-ated by adding the constraints
π΅πβ + π΅ππ β€ 1,β = 1 . . . πΏ β 3, π = β + 2 . . . πΏ, π = 1 . . . πΎ (5)
-
Advances in Operations Research 9
Figure 4: Nonsequential cutting-and-welding pattern for produc-ing three 11-meter pipes and three 3-meter pipes from seven single-size, 6-meter stocks.
and βπΏβ=1βπΎπ=1 π΅βπ = πΏ + πΎ β 1 to the objective functionβπΏβ=1 πβ(βπΎπ=1 π΅βπ) (see Section 4.2).The number of additional constraints increases quadrat-
ically in the number of stocks. If decision maker prefer-ences are ignored, the program generates feasible sequentialcutting-and-welding patterns with moderate running timesfor moderate Ks and possibly large running times for largeKs (see column G4 in Table 2). If decision maker preferencesare taken into consideration, the program generates mini-mal sequential cutting-and-welding patterns with runningtimes varying from (milli)seconds to hours (see pattern J inTable 2).
4.4.2. Reformulating the Pattern Conversion Problem forSequential Cutting-and-Welding Patterns and Decision MakerPreferences. For single-size stocks (6 meters, as in the casewhich motivates this article, or different), sequential cutting-and-welding patterns may also be generated by formulatingan assignment-like problem, a permutation of pipes, whereπ₯βπ β {0, 1} represents the place of pipe β in the permutationand where the objective function is arbitrary, because gener-ating one sequential pattern is sufficient:
πΏββ=1
π₯βπ = 1 π = 1 . . . πΏπΏβπ=1
π₯βπ = 1 β = 1 . . . πΏπβ1βπ=1
πΏββ=1
π βπ₯βπ β₯ ππ6000 π = 1 . . . πΏπβπ=1
πΏββ=1
π βπ₯βπ β€ (ππ + 2) 6000 π = 1 . . . πΏπ₯βπ β {0; 1} β = 1 . . . πΏ, π = 1 . . . πΏππ β N π = 1 . . . πΏ.
(6)
Theβπβ1π=1 βπΏβ=1 π βπ₯βπ β₯ ππ6000 constraints ensure that theaggregate size of the first π β 1 pipes in the permutation is
at least ππ6000; in other words, there are at least ππ stocksfor the first π β 1 pipes. The βππ=1βπΏβ=1 π βπ₯βπ β€ (ππ + 2)6000constraints ensure that the aggregate size of the first π pipesin the permutation is (ππ+2)6000 at themost; in other words,as well as the leftover segment from stock ππ, an additionalstock may be used for producing pipe π. These two setsof constraints ensure that pipe π + 1 in the permutationhas one weld at the most and that the program generates aminimal sequential cutting-and-welding pattern, albeit onewhich disregards decision maker preferences (see column S1in Table 2). The program running times are moderate forall patterns with the exception of J, which does not have aminimal integer solution.
Decision maker preferences may be taken into consid-eration, but the pattern conversion problem formulated as amixed integer linear program is more cumbersome; ifππ β{0; 1} is the binary variable indicatingwhether the πth pipe inthe permutation is welded or not andπ€β β {0; 1} is the binaryvariable which shows whether the βth pipe in the originalindexing is welded or not, then
πΏββ=1
πβπ€β β mins.t.: πΏβ
β=1
π₯βπ = 1 π = 1 . . . πΏπΏβπ=1
π₯βπ = 1 β = 1 . . . πΏπβ1βπ=1
πΏββ=1
π βπ₯βπ β₯ ππ6000 π = 1 . . . πΏπβπ=1
πΏββ=1
π βπ₯βπ β€ (ππ + 1) 6000 +ππ6000π = 1 . . . πΏ
πΏββ=1
πβ = πΎ β 1π₯βπ +ππ β€ π€β + 1 β = 1 . . . πΏ, π = 1 . . . πΏπ₯βπ β {0; 1} β = 1 . . . πΏ, π = 1 . . . πΏππ β N π = 1 . . . πΏππ β {0; 1} π = 1 . . . πΏπ€π β {0; 1} π = 1 . . . πΏ
(7)
Theβππ=1βπΏβ=1 π βπ₯βπ β€ (ππ + 1)6000 +ππ6000 constraintensures that pipe ππ in the permutation is welded ifππ = 1.The βoriginalβ index of pipe ππ is unknown, but the π₯βπ+ππ β€π€β + 1 constraint ensures that pipe β is welded if pipe β isin the πth place in the permutation and the πth pipe in thepermutation is welded.
-
10 Advances in Operations Research
Unfortunately, the algorithm running times turn out to belarge, longer than seconds just to generate a feasible solution(see column G5 in Table 2).
In summary, sequential cutting-and-welding patternsmay be generated in simple circumstances and possiblycomplex ones too, while decision maker preferences maybe taken into account in simple circumstances, but not incomplex ones.
5. Results and Discussion
The algorithm suggested in this article (see Section 4.1)was applied to the case of a Hungarian manufacturer offixed firefighting systems (see Section 4.4). The companyuses single-size, 6-meter stocks (see Table 1 in Section 1for illustrative pipe sizes and quantities), the stock price isπ1, and the welding price ππ€ is generally low. However,increases in demand lead to increases in both welding timeand opportunity cost, which in turn lead to increases inwelding price.
The software usedβBarCut (version 2.1)βwas suitablefor both single- and multiple-size stocks. The number ofstocks per cutting-and-welding patternwas limited toπΎ = 10and, to reflect the opportunity cost,ππ€ was set as a percentage(90%, 49%, and 30%) of π1, assumed invariable; Tables 3, 4,and 5 show theminimal cutting-and-welding patterns for thethree welding prices.
When ππ€ is high (ππ€ = 0.9π1), welding is used only wheninevitable, for pipes longer than 6 meters. To generate theoptimal equivalent cutting plan, the program running time is8.45 seconds and involves 30 stocks and 4 welds (see Table 3).
When ππ€ is moderate (ππ€ = 0.49π1), welding is used lessrestrictively. To generate the optimal equivalent cutting plan,the program running time is 8.47 seconds and involves 29stocks and 6 welds (see Table 4).
When ππ€ is low (ππ€ = 0.3π1), welding is used moreliberally. To generate the optimal equivalent cutting plan,the program running time is 8.76 seconds and involves 27stocks and 12 welds (see Table 5). Since the aggregate lengthof the leftover segments is just 5.034 meters, lower weldingprices and further investigationβwith ππ€ < 0.3π1βwouldbe useless.
Optimal equivalent cutting plans need to be convertedto minimal cutting-and-welding patterns. Since all the pipesin these optimal equivalent cutting plans were either shorterthan 6meters or involved only two stocks, the conversions areeasy; more complex cases were analysed in Sections 4.2 and4.4.
6. Conclusions
Due to its practical relevance across the entire industrialspectrum, the cutting-stock problem continues to be studiedintensively in operations research and to generate sophisti-cated algorithms solving it. Some of these were translated intoopen-source or commercial programs, catering to the diverseuser needs and circumstances, at various prices and undervarious terms and conditions.
Table 3: Minimal cutting-and-welding patterns for ππ€ = 0.9π1.Pattern Stockquantities
Weldnumbers Pipe sizes Repeats
1 1 0 2000+4000 12 1 0 4500 23 1 0 4680 114 1 0 5000 15 1 0 5096 16 1 0 5250 17 1 0 5260 18 1 0 5500 19 1 0 5660 110 1 0 6000 211 2 1 7200+4500 212 2 1 7000+4660 2
Table 4: Minimal cutting-and-welding patterns for ππ€ = 0.49π1.Pattern Stockquantities
Weldnumbers Pipe sizes Repeats
1 1 0 5260 12 1 0 2000+4000 13 3 2 4500+4500+4500+4500 14 2 1 4680+7200 25 2 1 5000+7000 16 1 0 4680 87 1 0 5096 18 1 0 5500 19 1 0 5660 110 1 0 6000 211 2 1 4680+7000 112 1 0 5250 113 1 0 4660 2
Since the first cutting-stock problems studied in opera-tions research were one-dimensional, increasing dimension-ality and diversifying methodology became natural exten-sions.This article is one of only a handful to focus on the effectof welding on the one-dimensional cutting-stock probleminstead; it is also the first of its kind to limit the number ofwelds to one, to conform to best practice safety requirementsfor fixed firefighting systems in the construction industry.
This article defined the cutting-stock problem with weld-ing (see Section 3.2) and argued that it can be converted toan equivalent cutting-stock problemwithmultiple-size stocks(see Section 4.1), a problem that can be solved with existingalgorithms and software (see Section 3.1). For probablythe first time in the literature, it also defined the patternconversion problem, whereby cutting patterns are convertedto cutting-and-welding patterns as a mixed integer linearprogram.
-
Advances in Operations Research 11
Table 5: Minimal cutting-and-welding patterns for ππ€ = 0.3π1.Pattern Stock quantities Weld numbers Pipe sizes Repeats1 1 0 6000 22 2 1 4680+7200 23 1 0 2000+4000 14 1 0 5660 15 1 0 5500 16 1 0 5260 17 1 0 5250 18 2 1 5000+7000 19 4 3 4660+4660+4680+4680+4680 110 2 1 4680+7000 111 3 2 4500+4500+4500+4500 112 4 3 4680+4680+4680+4680+4680 113 1 0 5096 1
Optimal solutions to the bounded cutting-stock problemswith welding were investigated with the algorithm suggestedin this article and constraints were examined (see Section 4.1).The implications for unbound cutting-stock problems withwelding could not be ascertained (see Section 4.3); however,the algorithmwas found to generate a heuristic solution at theleast (see Section 4.4).
Converting equivalent cutting patterns to cutting-and-welding patterns is fundamental to the algorithm suggested inthis article. For sequential cutting-and-welding patterns anddecision maker preferences, this pattern conversion problemwas (re)formulated as a mixed integer linear program (seeSection 4.4.2).
The cutting-stock problem with welding studied in thisarticle was one-dimensional; increasing dimensionality isa natural extension. However, two- and three-dimensionalcutting-stock problems with welding present further com-plexities which require further research and algorithms dif-ferent than the one suggested in this article; more likely thannot, large instances cannot be solved to optimality.
Data Availability
The data used to support the findings of this study areincluded within the article.
Disclosure
Preliminary version has been presented at 20th EuropeanConference onMathematics for Industry and at EURO/ALIO2018, International Conference on Applied CombinatorialOptimization.
Conflicts of Interest
The author declares that there are no conflicts of interestregarding the publication of this paper.
Acknowledgments
Kolos AΜgoston acknowledges the support of the HungarianAcademy of Sciences under its Cooperation of ExcellencesGrant (KEP-6/2017).The author is very grateful to colleaguesat the Department of Operational Research and ActuarialSciences, Corvinus University of Budapest, for relevant com-ments and suggestions on the earliest draft, to PeΜter BiroΜand AΜgnes Cseh at the Mechanism Design Research Unit,Institute of Economics, Centre for Economic and RegionalStudies, Hungarian Academy of Sciences, for precious com-ments and suggestions on another intermediary draft, andto Anamaria M. Cristescu-Martin and MaΜrton Benedek, foreditorial assistance with many drafts since.
References
[1] L. V. Kantorovich, Mathematical Methods of Organizing andPlanning Production, vol. 6, Publication House, Leningrad StateUniversity, Leningrad, Russia, 1939.
[2] L. V. Kantorovich, βMathematical methods of organizing andplanning production,β Management Science, vol. 6, no. 4, pp.366β422, 1960.
[3] P. C. Gilmore and R. E. Gomory, βA linear programmingapproach to the cutting-stock problem,β Operations Research,vol. 9, no. 6, pp. 849β859, 1961.
[4] P. C. Gilmore and R. E. Gomory, βA linear programmingapproach to the cutting-stock problemβpart ii,β OperationsResearch, vol. 11, no. 6, pp. 863β888, 1963.
[5] P. E. Sweeney and E. R. Paternoster, βCutting and packingproblems: a categorized, application-orientated research bibli-ography,β Journal of the Operational Research Society, vol. 43, no.7, pp. 691β706, 1992.
[6] G. M. Roodman, βNear-optimal solutions to one-dimensionalcutting stock problems,β Computers & Operations Research, vol.13, no. 6, pp. 713β719, 1986.
[7] G. WaΜscher and T. Gau, βHeuristics for the integer one-dimensional cutting stock problem: a computational study,βORSpectrum, vol. 18, no. 3, pp. 131β144, 1996.
-
12 Advances in Operations Research
[8] M. Gradisar and P. Trkman, βA combined approach to thesolution to the general one-dimensional cutting stock problem,βComputers & Operations Research, vol. 32, no. 7, pp. 1793β1807,2005.
[9] H. Dyckhoff, βA new linear programming approach to thecutting stock problem,β Operations Research. The Journal of theOperations Research Society of America, vol. 29, no. 6, pp. 1092β1104, 1981.
[10] G. Scheithauer and J. Terno, βA branchbound algorithm forsolving one-dimensional cutting stock problems exactly,βAppli-cationes Mathematicae, vol. 23, no. 2, pp. 151β167, 1995.
[11] P. H. Vance, C. Barnhart, E. L. Johnson, and G. L. Nemhauser,βSolving binary cutting stock problems by column genera-tion and branch-and-bound,β Computational Optimization andApplications, vol. 3, no. 2, pp. 111β130, 1994.
[12] F. Vanderbeck, βComputational study of a column generationalgorithm for bin packing and cutting stock problems,βMathe-matical Programming, vol. 86, no. 3, pp. 565β594, 1999.
[13] G. Belov andG. Scheithauer, βAbranch-and-cut-and-price algo-rithm for one-dimensional stock cutting and two-dimensionaltwo-stage cutting,β European Journal of Operational Research,vol. 171, no. 1, pp. 85β106, 2006.
[14] G. WaΜscher, H. HauΓner, and H. Schumann, βAn improvedtypology of cutting and packing problems,β European Journalof Operational Research, vol. 183, no. 3, pp. 1109β1130, 2007.
[15] J. Levine and F. Ducatelle, βAnt colony optimization and localsearch for bin packing and cutting stock problems,β Journal ofthe Operational Research Society, vol. 55, no. 7, pp. 705β716,2004.
[16] S. A. Araujo, A. A. Constantino, and K. C. Poldi, βAn evolution-ary algorithm for the one-dimensional cutting stock problem,βInternational Transactions in Operational Research, vol. 18, no.1, pp. 115β127, 2011.
[17] Y. Cui and Y. Yang, βA heuristic for the one-dimensionalcutting stock problem with usable leftover,β European Journalof Operational Research, vol. 204, no. 2, pp. 245β250, 2010.
[18] C. H. Cheng, B. R. Feiring, and T. C. E. Cheng, βThe cuttingstock problem - a survey,β International Journal of ProductionEconomics, vol. 36, no. 3, pp. 291β305, 1994.
[19] C. Goulimis, βOptimal solutions for the cutting stock problem,βEuropean Journal of Operational Research, vol. 44, no. 2, pp. 197β208, 1990.
[20] E. J. Zak, βThe skiving stock problem as a counterpart of the cut-ting stock problem,β International Transactions in OperationalResearch, vol. 10, no. 6, pp. 637β650, 2003.
[21] J. Martinovic and G. Scheithauer, βInteger linear programmingmodels for the skiving stock problem,β European Journal ofOperational Research, vol. 251, no. 2, pp. 356β368, 2016.
[22] D. Tanir, O. Ugurlu, A. Guler, and U. Nuriyev, βOne-dimensional cutting stock problem with divisible items,β 2016.
-
Hindawiwww.hindawi.com Volume 2018
MathematicsJournal of
Hindawiwww.hindawi.com Volume 2018
Mathematical Problems in Engineering
Applied MathematicsJournal of
Hindawiwww.hindawi.com Volume 2018
Probability and StatisticsHindawiwww.hindawi.com Volume 2018
Journal of
Hindawiwww.hindawi.com Volume 2018
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawiwww.hindawi.com Volume 2018
OptimizationJournal of
Hindawiwww.hindawi.com Volume 2018
Hindawiwww.hindawi.com Volume 2018
Engineering Mathematics
International Journal of
Hindawiwww.hindawi.com Volume 2018
Operations ResearchAdvances in
Journal of
Hindawiwww.hindawi.com Volume 2018
Function SpacesAbstract and Applied AnalysisHindawiwww.hindawi.com Volume 2018
International Journal of Mathematics and Mathematical Sciences
Hindawiwww.hindawi.com Volume 2018
Hindawi Publishing Corporation http://www.hindawi.com Volume 2013Hindawiwww.hindawi.com
The Scientific World Journal
Volume 2018
Hindawiwww.hindawi.com Volume 2018Volume 2018
Numerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisAdvances inAdvances in Discrete Dynamics in
Nature and SocietyHindawiwww.hindawi.com Volume 2018
Hindawiwww.hindawi.com
DiοΏ½erential EquationsInternational Journal of
Volume 2018
Hindawiwww.hindawi.com Volume 2018
Decision SciencesAdvances in
Hindawiwww.hindawi.com Volume 2018
AnalysisInternational Journal of
Hindawiwww.hindawi.com Volume 2018
Stochastic AnalysisInternational Journal of
Submit your manuscripts atwww.hindawi.com
https://www.hindawi.com/journals/jmath/https://www.hindawi.com/journals/mpe/https://www.hindawi.com/journals/jam/https://www.hindawi.com/journals/jps/https://www.hindawi.com/journals/amp/https://www.hindawi.com/journals/jca/https://www.hindawi.com/journals/jopti/https://www.hindawi.com/journals/ijem/https://www.hindawi.com/journals/aor/https://www.hindawi.com/journals/jfs/https://www.hindawi.com/journals/aaa/https://www.hindawi.com/journals/ijmms/https://www.hindawi.com/journals/tswj/https://www.hindawi.com/journals/ana/https://www.hindawi.com/journals/ddns/https://www.hindawi.com/journals/ijde/https://www.hindawi.com/journals/ads/https://www.hindawi.com/journals/ijanal/https://www.hindawi.com/journals/ijsa/https://www.hindawi.com/https://www.hindawi.com/