CHAPTER 6 Network Models ‘These is « multitude of operations research situations that can be modeled and solved «as networks (nodes connected by biauches). Some recent surveys report that as much 870% of the real-world mathematical programming probleuss van be represented bY network-related models. The following list ilustrates possible applications of networks 1, Design of an offshore natural gas pipeline network connecting wellheads in the Gulf of Mexico to an inshore delivery point. The objective of the model is to min- imize Une cost of constructing the pipeline. 2. Determination of the shortest route between two cities in a network of roads. 3. Determination of the maximum capacity (in tons per year) of a coal slurry pipeline network joining the cual imines in Wyoming with the power plants in Houston. (Slurry pipelines transport coal by pumping water through specially designed pipes) 4. Determination of the minimum-cost flow schedule from oil fields to refineries through a pipeline network. 5. Determination of the time schedule (start and completian dates) for the activi ties of a construction project. ‘The solution of these situations, and others like it, is accomplished through a variety of nctwork optimization algorithms. This chapter will present five of these algorithms. 1. Minimal spanning tree (situation 1) 2, Shortest-route algorithm (situation 2) 3. Maximum flow algorithm (situation 3) 4. Minimum-cost capacitated network algorithm (situation 4) 5. Critical path (CPM) algorithm (situation 5) 213214 Chapter6 Network Models 61 FIGURE 6.2 O G C G 3 . e) Examples ofa tee and. spanning tree 4 yy ® javen the network in Figure 6.1 ‘The situations for which these algorithms apply can also be formulated and solved as explicit linear programs. However, the proposed network-based algorithuus are more efficient than the simnplex method, NETWORK DEFINITIONS ‘A network consists of a set of nodes linked by ares (or branches), The notation for describing a network is (N, A), where Nis the set of nodes, and A is the set of arcs. As an illustration, the network in Figure 6.1 is described as N= (1,2, A = {012.0.3,2.3).25),.8.4,8,9,(42,4.5) FIGURE 6.1 Example of (N,A) network ing fl \4) Associated with each network is some type of flow (e.., oil produets flow in a pipeline and automobile traffic flows on highways). In general, the flow in a network is limited by the capacity of its ares, which may be finite or infinite. ‘An arc is said to be directed or oriented if it allows positive flow in one direction and zero flow in the opposite direction. A directed network has all directed ares. ‘A path is a sequence of distinct arcs that join two nodes through other nodes regardless of the direction of flow in each are. A path forms a eyele if it connects a node to itself through other nodes For example, in Figure 6.1, ares (2,3), (3.5),and (5.2) form loop. A cycle is directed if it consists of a directed path; e., (2,3), (3,4), and (4,2) in Figure 6.1 A connected network is such that every two distinet nodes are linked by at least one path. The network in Figure 6.1 demonstrates this type of network. A tree is a connected network that may involve Uuly a sulvet of all the nudes of the network with no cycles allowed, and a spanning tree is a tree that links all the nodes of the notwork, algo with no cycles allowed Fignre 67 provides examples of a tree and a spanning tree for the network in Figure 6.1. Tee Spanning teejon for es. AS ow ina work is rection : nodes sa node 2) form (4,2) in at least ree is 0 jetwork, s of the e and a 6.2 Minimal Spanning Tree Algorithm 215 PROBLEM SET 6.10, 1. For each network in Figure 6.3 determine (a) a path, (b) a cycle. (c) a directed eycle,(d) a tree, and (e) a spanning tree, FIGURE 6.3 Networks for Problem 1,Set 6.14 «wo Determine the sets V and A for the networks in Figure 6.3. Draw the network defined by N= {1.2.3.4.5.6} A> ((12K0.5.23124.8.5)184)(4.3944,0)15.2),5.6} Consider eight equal squares arranged in three rows, with two squares in the first row, {our in the second, and two in the third. The squares of each row are arranged symmetri- cally about the vertical axis Its desired to fill the squates with distinet numbers in the range 1, 2,...,and 880 that no two adjacent vertical, horizontal, or diagonal squares hold consecutive numbers, Use network representation as a vehicle to find the solution in a systematic way, + Three inmates escorted by 3 guards must be transported by boat from San Francisco to the Aleatraz penitentiary stand to serve their sentences. The boat cannot transfer mute than two persons in either direction. The inmates are certain to overpower the guards if they outnumber them at any time. Develop a network model that designs the boat trips in ‘a manner that ensures a safe transter ofthe inmates, Assume that the inmates will not flee if given a chance, MINIMAL SPANNING TREE ALGORITIIM ‘The minimal spanning tree algorithm deals with linking the nodes of a network, ircetly or indiveetl, using tie shortest length of connecting branches. typical appli: cation occurs in the construction of payed roads dat link several towns. The road between two towns may pass through one or more other towns, The most ecuuvinical design of the road system calls for minimizing the total miles of paved roads, a result, that is achieved by implementing the minimal spanning tree algorithm, ‘The steps of the procedure are given as follows. Let N = {1,2, ...,} be the set of nodes of the network and detine et of nodes that have been permanently connected at iteration k jet of nodes as yet to be connected permanently216 Chapter Network Models FIGURE 6. Step0. Set Cy = Gand Cy = N Step 1. Start with any node, j, in the unconnected set Cy and set C; = {i}, which ren- ders C, = N ~ {i}.Setk = 2 General Step k. Select a node, ’,in the unconnected set C,.. that yields the shortest are toa node in the connected set C,... Link j" permanently to C,.; and remove it from C1, that is, C= Cr + FRC = Gen - I If the set of unconnected nodes, C,, is empty, stop. Otherwise, set k = k + 1 and repeat the step, Example 62-1 Midwest TV Cable Company is in the process of providing cable service to five new housing development areas Figure 64 depiets possible TV linkages among the five areas. The cable miles are shown on each arc. Determine the most economical cable network, The algorithm starts at node 1 (any other node will Ju as well), which gives C, = OG, = 2,3.4,5.6} The iterations of the algorithm are summarized in Figure 6.5. The thin arcs provide all the candidate links between C and C. The thick branches represent the permanent links among the nodes of the connected set C, and tlie dashed branch represents the new (permanent) link added at each iteration. For example, in iteration 1, branch (1.2) is the shortest link (= I mile) among all the candidate branches from node 1 to nodes 2, 3, 4, and 5 of the unconnected set C,. Hence, link (1,2) is made permanent and j’ = 2,which yields, 12,0) = (3.4.5.6) The solution is given by the minimal spanning tree shown in iteration 6 of Figure 65.The resulting minimum cable miles needed to provide the desired cable service are 143444345 = 16 miles A Cable connections for Midwest TV Cable Company}. which ren- the shortest C.-.and etk=k+1 © 10 five new mong the five nomial cable cs provide all se permanent epresents the jon 1, branch om node 1 to je permanent va 6 of Figure le service are re 6.2 Minimal Spanning Tree Algarithm 247 FIGURE 6.5 Solution iterations Iteration S eration 6 for Midwest TV (Minimal spanning tree) Cable Company You can use TORA to generate the iterations of the minimal spanning ue-Fivit ‘aki ens, select NetWork madelA <> Wninat spaining eae. Next, from souve/woozey menu, select Solve problen => Gs to eutgut screen. In the output screen, select a Stazting sede and then use Ne iteration or AI1 iee¥ations to generate the succes- swe iterations. You can restart the iterations by selecting a new starting node. Figure 66 gives TORA output for Example 6.2-1 (file ch6 LoraMinSpanEx6-2-L.txt).218 Chapter Network Models SI Her ak luis a ICI ees TOR FIGURE 6.6 Output ofthe minimal spanning tee of Example 62-1 PROBLEM SET 6.28 1. Solve Example 6.2-1 starting at node S (instead of node 1), and show that the algorithm produces the same solution, 2. Determine the minimal spanning tree of the network of Example 62-1 under each of the following separate conditions (a) Nodes 5 and 6 are linked by a 2-mile cable. (b)_ Nodes 2 and 5 cannot be linked. (©) Nodes 2 and 6 are linked by a 4-mile cable. {@) The cable between nodes 1 aud 2 iy 8 miles Kou, (©) Nous 3 and 5 arc linked by a 2-mile eable. (f) Nodo 2 cannot be linked directly to nodes 3 and. 3. In intermodal transportation, loaded truck trailere are chipped hetween ra nals by placing the trailer on special flatbed carts. Figure 6.7 shows the location of the ‘main railroad terminals in the United States and the existing railroad tracks. The objec- tive is to decide which tracks should be “revitalized” to handle the intermodal traffic. In particular the Los Angeles (LA) terminal must be linked direetly to Chicago (CH) to accommodate expected heavy traffic. Other than that all the remaining terminals can be linked, directly or indirectly, such that the total length (in miles) of the selected tracks is, ‘minimized, Determine the segments ofthe railroad tracks that must be included in the revitalization program. Iran tormi-jeorithm ch of the termi of the 32 objec- aific. In CH) to als can be tracks is inthe 6.2 Minimal Spanning Tee Algorithm 219 FIGURE 6.7 ‘otwork for Problem 3, Set 6:24 4. Figure 0.8 gives the mileage of the feasible inks connecting nine offshore natural gas ‘wsllheade with an inshore delivery point. Because the lvativu of wellea 1 is the closest to shore, its equipped with sufficient pumping and ctorage eapacity to pump the output of the remaining eight wells to the delivery point. Determine the minimum pipeline net- work that links the wellheads to the delivery point. Botner pot CURES Network for Problem 4 Sit 628 In Figure 6.8 of Problem 4, suppose that the wellheads can be divided into two groups ‘depending on gas pressure: a high-pressure group that includes wells 2,3, 4,and 6: and a low pressure group that includes wells 5,7, 8,and 9. Because of pressure difference, well- hheads from the two groups cannot he linked. At the same time, both groups must be con nected to the delivery point through wellhead 1, Determine the minimum pipeline network for this situation, {electro produces 15 electronic parts on 10 machines. The company wants to group the ‘machines into cells desiguted (o minimize the “dissimilarities” among the parts processed220. Chapter Network Models in each cell. A measure of “dissimilarity,” d,, among the parts processed on machines { san va be eapressed as tol oe my ‘where ny is the number of part shared between machines and j, and my is the number of i parts that are used by either machine ior j only ‘The following table assigns the parts to machines Machine Assigned parts 1 16 2 23,7.89)1213.15 3 351014 4 27a 5 X51011.14 6 Lass10 BOBS 9 410 Ww saya 19 (@) Feprese the prohlem asa network model (b) Show that the determination of the cells ean be based on the minimal spanning tree solution, (©) For the data given in the pret eding table, construct the two- and three-cell solutions 6.3 SHORTEST-ROUTE PROBLEM The shortest-route problem determines the shortest route between a source and desti- nation in a transportation network. Other situations can be represented by the same model as illustrated by the following examples. 6.3.1. Examples of the Shortest-Route Applications Example 6.3-1 (Equipment Replacement) RentCar is developing a replacement plan for its car fleet for a 4-year planning hori- zon that starts January 1.2001, and terminates December 31, 2004. At the start of each year, a decision is made as to whether a car should be kept in operation or replaced. A car must be in service a minimum of | year and a maximum of 3 years. The following table provides the replacement cost as a function of the year a car is acquired and the number af years in aperation, Replacement cost (§) forgiven years in operation aupment acquired tstartof 1 2 3 2003 3800 7100 2004 3900)d desti je same wg hori- of each aced. A owing and the 6.3. ShortestRoute Problem 221 ee Equipment replace =o tent protien asa shortest-route mode The problem can be formulated as a network in which nodes 1 to 5 represent the Mart of years 2001 to 2005. Ares from node 1 (year 2001) can reach only nodes 2, 3, and 4 because a car must be in operation between 1 and 3 years. The ates frum the other nodes can be interpreted similarly. The length of each arc equals the replacement cost. ‘The solution of the problem is equivalent to finding the shortest route between nodes Land 5 Figuse 6.9 sliows the resulting network. Using TORA,' the shortest route (shown by the thick path) is 1+ 3— 5. The colution means that a car acquired at the stait uf 2001 (node 1) must be replaced after 2 years at the start of 2003 (nadie 3). The replace ment car will then be kept in service until the end of 2004. The total cost ofthis replace ment policy is $12,900 (= $5400 + $7100), Example 6.3-2 (Most Reliable Route) 1. Q. Smart drives daily to work. Having just completed a course in network analysis, Smart is able to determine the shortest route to work. Unfortunately, the selected route is heavily patrolled by police, and with all the fines paid for speeding, the shor route may not be the best choice. Smart has thus decided to choose a route that maxi mizes the probability of nor being stopped by police. The network in Figure 6.10 shows the possible routes between home and work, ‘aud Ue associated probabilities of not being stopped on each segment. The probability of not being stopped on the way to work is the product of the probabilities assuciated with the successive segments of the selected ronte. For example, the probability of not, FIGURE 6.10 Most-reliable-route network mode! PY menu, select sive pes222° Chapter6 Network Models > 4598) CA sous 7 Sy) FIGURE 6.11 1 sani MD Most reliable ru “as representation as a shortest 3 Mm syute model A 8 receiving a fine on the route |» 3-5 > 7is 9 X 3 X 25 = .0675.Smart’s objective is to select the route that maximizes the probability of not being fined The problem can be formulated as a shortest-route model by using a logarithmic transformation that converts the produet probability into the sum of the logarithms of probabilities—that is if pu = p. X pa X ... X pu is the probability of not being stopped, then log p,, = log p, + log py + ... + log py Mathematically, the maximization of p,« is equivalent to the maximization of log pus Because log yy, $ 0, the maximization of lug py is in tui, equivalent the minimization of — log p,,. Using this transformation, the individual probabilities p, in Figure 6.10 are replaced with ~ loz p, for all jin the network, thus yielding the shortest-route network in Figure 6.11 Using TORA, nodes I, 3,5, and 7 define the shortest route in Figure 6.11 with a corresponding “length” of 1.1707 (= —log p,;). Thus, the maximum probability of not being stopped is py, = .0675. Example 63-3. (Three-Jug Puzzle) An 8-gallon jug is filled with fluid. Given two empty 5- and 3-gallon jugs, we want to divide the ¥ gallons of tluid into two equal parts using the three jugs. No other measur- ing devices are allowed. What is the smallest number of pourings needed to achieve this result? You probably can guess the solution of this puzzle. Nevertheless. the solution process can be systematized by representing the problem as a shortest-route problem ‘A node is defined to represent the amount ot tluid in the b-,5-, and 3-gallon jugs, respectively. This means that the network starts with node (8,0,0) and terminates with the desired solution node (4, 4,0) A new node is generated from the curren pouring fluid from one jug into another Figure 6.12 shows different routes that lead from start node (8, 0,0) to end node (4, 4,0). The are between two successive nodes represents a single pouring, and hence can be assumed to have a length of 1 unit. The problem reduces to determining the shortest route between node (8,0,0) and node (4.4.0). ‘The optimal solution, given by the bottom path in Figure 6.12, requires 7 pourings. nade by