You are on page 1of 31

MIDDLE EAST TECHNICAL UNIVERSITY

I E 3 9 8 S Y S T E M S T H I N K I N G

Term Project:
Scheduling a Fleet of Road Tankers

Burcu Yzak - 1627884 Fato lbi - 1535459 Onur Ylmaz - 1627868

May 2011, Ankara

Table of Contents
PAGE Table of Contents .............................................................................................................. 1. INTRODUCTION .. 1.1. Project .............................................................................................................. 1.2. Related Issues ........................................................................................................ 1.3. Approach ............................................................................................................ 1.4. Recommendations ................................................................................................ 2. REPORT .... 2.1. Statement of the Problem....................................................................................... 2.2. Analysis of Specific System...................................................................................... 2.3. Scale Decisions and Critiques.................................................................................. 2.4. Major Steps of Analysis and Findings...................................................................... 2.5. Alternative Actions.................................................................................................. 1 2 2 2 2 2 3 3 5 8 9 14

2.6. Recommendations................................................................................................... 16 3. CONCLUSION .................................................................................................................. 17 4. GLOSSARY . 5. APPENDIX .... 18 I

1. INTRODUCTION
1.1. Project
This project is related to planning the fleet size and its content; and scheduling the fleet of Transport Department of Asit Kimya in order to meet the next periods demand. This project is not only based on managerial trade-offs and decisions but also includes engineering approaches like planning and scheduling considering the limitations and objectives.

1.2. Related Issues


In this project, not all the operations of the Asit Kimya or Transport Department are considered. Scheduling the current fleet, changing the size and its content are the main concerns of this study; on the other hand, maintenance operations, purchasing operations and limitations of these operations are not considered in the concept of this study. In addition, assumptions are made on the unknown or unspecified details of the operations while considering the main structure of the problem situation.

1.3. Approach
The manager is not willing to clean the tankers since it is hazardous to cleaning workers. He finds the cleaning decision reasonable only when he will need to buy a new tanker in order to meet the demand of the next year.Thus, his approach to cleaning and purchasing decisions will be reflected to the solution approach by using step by step approach. At each step, cleaning or buying one more additional tanker options will be compared. In addition to these, alternative actions will also be introduced.

1.4. Recommendations
It is recommended that Transport Department of Asit Kimya should proceed with a preliminary analysis of the transportation operations in detail by its costs and critical requirements. The analysis would develop a model for finding optimal scheduling scheme as well as optimal fleet size, i.e. exact numbers specifically for every type of tanker, via indicating number of cleanings at the beginning of the year and new tankers to be bought.

According to this model, reliable estimates of the potential savings in operating costs should be computed in order to justifying the results of the model.

In the next parts of the report, firstly, problem situation is described and then specific system which is going to be studied is given. Following these, boundary of the system is determined by declaring assumptions on the system. Then, major steps of analyzing this system are shown in detail and findings and comments on these findings given in the subsequent section. Considering these findings, alternative actions that can be taken are mentioned and the final recommendations are made in the last section.

2. REPORT
2.1. Statement of the Problem
In this part, problem which will be solved in the next parts will be defined with presenting different aspects and different roles of people which are involved in the problem.

Firstly, dividing the problem into smaller subsets will make it easier to analyze and realize their effects on each other and the whole.

Since this is a business environment, whether to implement the solutions provided will be the decision of the manager of the Transport Department of Asit Kimya and thereafter he will be mentioned shortly as manager. Since this project aims to solve a problem, there will be an objective or some objectives. In this manner, the managers objective is to operate the Transport Department successfully by the means of its measurable and immeasurable aspects.

In order to control how well the main objective is reached, this one objective can be divided into more specific goals on different areas which would provide rather small environment to consider for a more focused study. First of these goals is the minimizing the total costs which is the sum of transportation costs, procurement costs and operating costs. Second is to minimize number of cleanings and damage of this cleanings on the cleaning workers. As the third goal, it is aimed to minimize the spare time of trucks throughout the year. For the last, it is aimed to minimize the possible threats to public safety caused by
3

accidents etc. These specified goals will determine how well the solutions satisfy the main objective.

These objective and goals should be measured by some means, in order to compare different actions. These means could include numerical measurements like total transportation cost, procurement cost, operating cost; or in short overall total costs, or rates and counts such as number of cleanings and proportion of the spare time to the all available time. On the other hand there are some aspects which are difficult to assign numeric values but reveals the performance of the solutions, like amount of damage given to cleaning workers and amount of damage caused by tanker accidents

In order to derive a proper solution to the problem, there shall be different actions to be taken in this problem environment, and those can be listed as the following; determining number of tankers to buy and sell, number of compartments to clean and finally allocation of tankers to route. These actions, individually or as a group, will shape the solution provided to this problem.

Finally, there is an environment in which this problem is emerged and needed to be solved. Therefore, context of the problem which affects the situation must be presented. There are some direct relationships that no one has control on them, such as cleaning operations effects on health problems; also, some legal and operational limitations which cannot be changed, like the limitation of 16.5 tons for any type of material stated by law and minimum/maximum level of materials to be delivered to a depot due to the capacities of these depots. In addition to these, costs of new tankers or salvage, taxes, insurance and transportation cost per kilometer are taken as given. Since the problem is about planning the succeeding year, and the manager stated so, demand forecasts of the next year are counted as completely reliable. Moreover, due to security issues, maintenance operations and their time requirements, which is 40 % of the total annual time, are also included in the problem environment.

Secondly, presenting the different roles of individuals which are included in the problem situation would make the problem statement more clear. The owner of this problem, who has the full responsibility of the possible consequences of this problem, is the manager of the Transportation Department. When the problem is solved, people who are going to execute
4

the decisions are the tanker drivers, cleaning and maintenance workers. In addition, there are the ones who would benefit or would be victim of the consequences of these implemented decisions. They are other depots which are in Asit Kimyas transportation network, cleaning and maintenance workers and publicity, due to the possible threats. This problem will be solved and recommendations will be provided by the analysts in OR Department of the Asit Kimya.

To conclude, in this part, problem is divided into parts and each part is described in detail. Following that, different roles of individuals are mentioned in order to provide a full explanation of the problem statement. In the following part, specific system which this problem is emerged will be explained.

2.2. Analysis of Specific System


In the previous section, the problem is stated and in this following section, system in which the problem is emerged is going to be studied and the point of view will be fixed for the next phases of providing solution. In this manner, if we consider the Transport Department as a black box then it will transform existing routes, tankers and workers; with the possible new additions/removals into a new route scheduling for the following year. In order to accomplish this transformation, some sub-components of the system; namely, cleaning system, maintenance system, fleet of tankers, factory and pre-determined routes; are used. Considering this system as a black box it is convenient to look out for some inputs and outputs. First of all, these inputs can be gathered into two subsets as the ones which are not in the control of the system and the ones which are going to be decided. The first set consists of demand forecast, number of on-hand tankers and their types, number of current workers and wages, costs of procurement, mileage cost, taxes and insurance, limits of other depots, security limits of carriage, maintenance time limit, current compartment allocation of tankers, existing routes and capacity of tankers and compartments. And the second set includes the inputs which could be decided; explicitly, number of new tankers, number of tankers salvaged, number of cleanings and capacity usage of compartments.
5

Besides these controllable and uncontrollable inputs, there should be some outputs of the system. Although, some of them are difficult to measure; these expected outputs of the system could be listed as; total cost (transportation cost, procurement cost, operating cost), damage given to cleaning workers, spare time of tankers, damage caused by tanker accidents and a new schedule of tankers.

All aspects of the system mentioned could be summarized by the given diagram which shows the narrow system of our interest specifically:

Diagram 2.2: Narrow system of interest

As summarized in the influence diagram below, controllable inputs are shown with rectangular and uncontrollable ones are shown with clouds. The different situations of the system, which are formed by taking these inputs and processing them, are indicated by circles. Outputs which are shown with ovals should be mentioned according to their importance in the objective of the system. This system is trying to accomplish its objective by minimizing total costs, number of cleaning operations, spare time of tankers, damage given to cleaning workers and damage caused by tanker accidents.
6

Diagram 2.3: Influence diagram

2.3. Scale Decisions and Critiques

Having settled the system to be studied in the last section, assumptions and critiques will be presented with their justifications before going any further in analyzing the system, in the following part of this report. First of all, as mentioned before, demand forecasts are taken as given for the planning of next years operations without any concern. As known, Transportation Department Manager implied that there are negligible deviations in the numbers when previous years forecasts considered. Therefore, it is reasonable to assume that forecasts are totally true for next year. Secondly, being a decision maker, Transportation Manager is presumed that has some authority limitations and relaxations. Although there are any clear evidence; Transportation Manager is counted as having the power of buying and selling tankers without any restrictions like smooth level of resources or monetary issues. On the other hand, manager has some restrictions, for instance; it is regarded as he has no chance to change the other depots load/unload demands and capacities. Thirdly, there are some assumptions about operations of the factory and its subdepartments. All considerations are made in an environment such that the workforce is fixed. That is because, any cost of hiring new drivers or workers for maintenance when new tankers are bought is given; in addition to these, also any cost for firing employees when the size of fleet is decreased is mentioned. Considering all of these, workforce is taken as constant for the following year or there would be no effect of changing the level of workforce for our concern. The second issue about operations is the currently used routes. Routes which are currently using will be used without any change in the following year. This assumption would not only will ease the analysis of the recommendations, but also, fix the focus of the study to more important aspects of the situation. Finally, since it is mentioned that, due to security issues, any change is considered on the maintenance checks, therefore, total available time of one tanker is taken as 5240 hours annually. Fourthly, there are some assumptions about types and assignments of tankers. It is assumed that when the manager wants to buy new tankers, the only available tankers are the ones which are mentioned in the question text, i.e. A, B, C, D; and there is no other option. In
8

addition, it is assumed that any type of tanker could be assigned to any of the routes without any restriction like road conditions, insurance restrictions etc. Fifthly, there are some immeasurable aspects which cannot be directly included in the cost calculations. For instance, there is no assigned cost of cleaning one tanker and more importantly there is no fixed cost of the damage given to the cleaning workers. Therefore, it is assumed that minimizing the number of cleanings could minimize both of these and it will satisfy the concerns of the manager. Finally, in this project it is tried to give a scheduling output which only includes assignment of one type of tanker to one route and this approach is undertaken in order to avoid combining or dividing routes throughout the next planning horizon. Since this approach of assignment seems to ignore idle time of the tankers by not assigning them to other routes in their idle times, and since spare time of the tankers are tried to be minimized, at all steps of analysis, idle time percentages are checked prior to making any recommendations. Considering these assumptions and decisions made to fix the scale of our view, our solution approach to the problem will be presented in the following part.

2.4. Major Steps of Analysis and Findings

First of all, the main thing that should be focused on, while constructing the model of the system, is the trade-off between the number of cleanings and the number of new tankers to be bought. If a tanker is cleaned, it will not affect the total cost considerably but the harm that will be given to the cleaning workers must be taken into consideration. On the other hand, if a new tanker is bought, only the high level of cost will be considered; however, the responsibility on the health of the cleaning workers is not so easy to consider. So the manager is willing to undertake the cleaning if he can see that it will save him from purchasing another vehicle to transport the forecast of the coming year. Due to the trade-off explained above, it is thought that solving the problem using one model will not be effective on finding a solution. Therefore, to solve the problem different models; in fact modified models, are used in each step. Note that the number of tankers allocated to routes, number of tankers cleaned and the number of new tankers bought are
9

defined as integer numbers in the models in order to find a logically valid solution. However, number of times a tanker completes its route is assumed to be a real number to make the solver of the problem, GAMS, solve efficiently. Moreover, salvaging the unnecessary tankers is also considered in each step. The major steps followed in order to reach the best solution to the problem are summarized below: a) First Step: In this step, the aim of the model is to be acquainted with the system and to see whether the tankers on hand will be sufficient to meet the demand of the next year without cleaning the existing ones. The model is constructed using the idea that depots are covered by routes. It means that if the demand of the depot is going to be satisfied then there must be at least one tanker which is assigned to the routes covering that depot. The idea behind this model is to assign types of tankers (A,B,C,D) to the routes. So an assumption made here is that this assignment is done at the beginning of a year and it does not change during the year. Decision variables are defined for each route and each tanker type indicating the number of tankers of given type assigned to the given route. In addition to these, there are also decision variables for each tanker type indicating the number of salvaged tankers and one more decision variable for number of times a tanker completes its given route. The model includes constraints mentioned by the manager of the company. First of all, there are demand satisfaction constraints to satisfy the demands of the next year. Moreover, maintenance constraints saying that 40% of a year a tanker should be in maintenance, legal restrictions obligating a delivery amount no more than 16,5 tons and the restrictions of the depots on the delivery amounts are also included. The details of the model and brief explanations of the constraints can be found in Appendix A. Findings: In this step, it is expected to find out whether or not the current situation is sufficient for the next periods. Solving the model explained in the first step of analysis, it is found that it is not possible to meet the demands of the next year without changing the number of on hand tankers. Therefore, the manager should take any actions like either existing tankers should be cleaned or the manager should buy new tankers. Details of the solution are given in Appendix B showing that the solution is infeasible.
10

b) Second Step: In the last part, it is shown that it is impossible to meet the demand of the next planning horizon with the current number of tankers. Since manager is willing to undertake cleaning only if it saves him from buying a new tanker, and current number of tankers is not enough, it is thought that, what would happen if the model gives full permission to cleaning operation? With this reasoning, firstly, new decision variables are added in order to determine whether or not a cleaning operation is undertaken, how many of tankers are cleaned and combining these with the current level of the tankers, number of tankers is updated. Secondly, some constraints related to change in the number of tankers due to cleaning operation are added and one more constraint related to assigning number of cleaned tankers is added. Thirdly, since there is no determined cost of the cleaning, and this step only checks whether cleaning is a solution to infeasibility of the first step, no cleaning related cost is added to objective function. Finally, it is thought that only cleaning between A-B types and C-D types considered since, for instance, A-D cleaning exceeds the 16,5 tons of legal carriage amount. Changes described above are made to the model of the first step and it is provided in the Appendix C. Findings: By solving the model of the second step, it is found that it is impossible to meet the demand of the next period even with the restrictions on cleanings are removed. This result means that there should be an increase in the total numbers of tankers because in this step all operationally feasible compartment combinations of the currently available trucks are tested. As given in the Appendix D, relaxing the considerations on cleaning will also yield an infeasible solution.

c) Third Step: In the last part, it is shown that it is impossible to meet the demand of the next planning horizon with the current number of tankers even if the cleaning operations are undertaken. Therefore, this step is focused on what would happen if we let manager buy any number of tankers which yields the minimum cost. This step will going to be a base step for the following steps, because this will provide a direction on the type and number of tankers to buy and trade-offs between new tankers and cleanings will be made on this direction.
11

With this reasoning, some changes are made to the model in the first step. First of all, number of tankers and their types are added to decision variables and this will determine the total number of tankers with their types. Secondly, a constraint added to ensure that the new bought tankers are added to the number of the related tankers. Thirdly, objective function is updated to include new purchase cost. Changes described above, which are made to the model in the first step, are given in the Appendix E in detail. Findings: In this step, it is expected to find the optimal number of trucks with the minimum total cost. Because, since in the last two steps yield infeasible solutions, in this step purchase is allowed with no number limitation. As given in the Appendix F, this step has a minimum cost level of $ 110.846,73 and route-tanker assignments with related numbers can be seen from the Table 2.5 below: R 1 A B C D TANKERS 3 1 1 1 1 1 1 1 2 Table 2.5.1: Information gathered from Appendix F 1 2 3 4 5 O 6 U 7 T E 8 1 S 9 10 11 12 1 3 6 4 TOTAL TANKERS

As tabulated above, result of this step shows that the minimum cost is attained when 4 new D type tankers are purchased and shown assignments are made. As mentioned in the last parts related step above, since this step creates a basis for the direction of the following steps, in the next steps trade-offs will be made considering this 4 new D type tankers. It is shown that it is impossible to meet the demand of the next planning horizon without any purchase. In addition, it is found that if all new purchases made on type D and no cleaning is made then the minimum cost will be attained. Therefore, in the next steps, purchasing one more D type tanker and cleaning trade-off will be compared. With this reasoning, in the next steps, 8000 of new purchase cost will be added to total cost function and on hand D type tanker will be increased by one in each step. Then it is checked whether or not the model is feasible without buying any additional tanker and cleaning the on hand tankers. This iterative controls are made for considering that we have one, two and three D types of tankers on hand, because if were to buy 4 D type tankers, considering any cleaning
12

will be unnecessary since the manager wants to undertake cleaning only if it saves him from buying any additional tanker.

d) Fourth Step: In this step, it is checked whether or not it is possible to purchase one D type tanker and undertake cleaning operations to meet demand. With this approach, changes made to model of the Second Step are provided in Appendix G. Findings: By solving this model, it is found impossible to meet demand with one additional D type tanker and letting cleaning operation if there is any need.

e) Fifth Step: In this step, it is checked whether or not it is possible to purchase two D type tankers and undertake cleaning operations to meet demand. With this approach, changes made to model of the Second Step are provided in Appendix H. Findings: By solving this model, it is found impossible to meet demand with two additional D type tankers and letting cleaning operation if there is any need.

F) Sixth Step: In this step, it is checked whether or not it is possible to purchase three D type tankers and undertake cleaning operations to meet demand. With this approach, changes made to model of the Second Step are provided in Appendix J. Findings: By solving this model, it is found impossible to meet demand with three additional D type tankers and letting cleaning operation if there is any need.

Considering the fourth, fifth and sixth steps, it is found that the manager cannot use cleaning to save himself from buying additional D type tankers.

13

2.5. Alternative Actions

As explained above, the first feasible solution to this problem is found in the third step in which the cleaning option is not considered and buying new tankers is taken into consideration only. The decision was to buy 4 new type D tankers. Then in order to reflect the trade-off between cleaning decision and buying decision, the model including cleaning in second step is modified to reach a feasible solution by increasing number of type D tankers on hand by one in each step up to three type D tankers. In these steps, it is found that cleaning cannot overcome the need for type D tankers. So it is concluded that the best solution to that problem is to buy 4 new type D tankers. However, some alternative scenarios are also found. While deciding on each scenario, the idea was to see whether other tanker types can be bought instead of type D tankers when the cleaning is also considered. To be able to compare each scenario easily all combinations of options are shown by using tables. Since buying a type B tanker will result in cleaning them and increasing the number of type A tankers due to larger demand for acidic than for caustic and larger acidic capacity of type A tankers than of type B tankers; while tabulating only purchase of type A tankers are combined with purchase of type C tankers. Each feasible scenario has the cost value and the number of cleaning done in that scenario. Moreover, some of the scenarios are cancelled since they include buying more than four tankers, which will directly be more costly than buying four new type D tankers.
Number of type A tankers bought 0 0 Number of type C tankers bought Fourth Step Infeasible 1 Infeasible 2 Infeasible Cost: 111.646,73 # of Cleanings: 3 (C type) 3 Cost: 116.204,48 # of Cleanings: 2 (C type) -

Infeasible

Infeasible Cost: 111.646,73 # of Cleanings:


3 (C type) 1 (B Type)

Infeasible Cost: 111.646,73 # of Cleanings: 3 (C type)

Table 2.5.2 : Number of A and C type tankers ara compared when number of on hand D type tankers is 1 14

Number of type A tankers bought 0 0 Number of type C tankers bought Fifth Step Infeasible Infeasible Cost: 111.646,73 # of Cleanings: 3 (C type) 1 Infeasible Cost: 111.646,73 # of Cleanings:
2 (C type) 1 (B Type)

2 Cost: 111.646,73 # of Cleanings: 2 (C type) -

3 -

2 3

Table 2.5.3 : Number of A and C type tankers ara compared when number of on hand D type tankers is 2

Number of type A tankers bought 0 0 Number of type C tankers bought Sixth Step Infeasible Cost: 111.646,73 # of Cleanings: 2 (C type) 1 Cost: 111.646,73 # of Cleanings: 2 (C type) 2 3 -

1 2 3

Table 2.5.4 : Number of A and C type tankers ara compared when number of on hand D type tankers is 3

Above tables show the options when one, two and three type D tankers are bought respectively. The options are elected by comparing the costs first, then number of cleanings done is considered and the ones having larger number of cleanings are omitted. Thereafter, options having larger caustic capacity are ignored since caustic demand is not high. Finally, two alternative actions are found which are shown in Table 2.5.5 below. GAMS outputs of the models for Option 1 and Option 2 are in Appendix K and L respectively.

15

Third Step Total Cost # of cleanings 110.846,73 -

Option 1 111.646,73 2 ( C type )

Option 2 111.646,73 2 ( C type ) 2 3 4 5

# of type A tankers 1 3 # of type B tankers 3 3 # of type C tankers 6 4 # of type D tankers 4 4 Table 2.5.5: Alternative Actions and Third Step Compared

2.6. Recommendations

As seen in Table 2.5.5 the cost value of buying four type D tankers is the optimal one. Therefore, it will be better to choose this option. Moreover, in these two alternative options, number of type A tankers are increased by buying new ones. However, buying type D tankers is always more meaningful than buying type A tankers. The reason for that is that type D tankers have extra caustic capacity compared to type A tankers. In addition to that, they have the same price, 8000TL.Therefore, buying four new type D tankers is again the optimal policy when looked from that point of view. As mentioned before, outputs of the system are schedule of the tankers which is given in Appendix F for the optimal case, number of cleanings done which is zero, total cost including procurement, operating and transportation cost minus salvages, which is found to be 110,846.73TL. In addition to these, Table 2.6.1 below shows the spare time percentages of a tanker assigned to each route for the optimal policy.

% spare time of tankers

Route 1 0 Route 7

Route 2 0 Route 8 31

Route 3 45 Route 9 0

Route 4 36 Route 10 69

Route 5 66 Route 11 90

Route 6 0 Route 12 96

% spare time of tankers

18

Table 2.6.1: Spare time of tankers

16

As it can be seen from the above table some tankers assigned to specific routes have a large amount of spare time. The reason for that is the assumption made which says that if a tanker is assigned to a route at the beginning of the year, then it will not be assigned to another route. If this assumption is considered again, then the number of tankers needed to meet the requirements of the following year could be less than 14 since the tankers would be assigned to different routes at their spare times.

3. CONCLUSION
In this project, scheduling the fleet of Transport Department of Asit Kimya is analyzed. The problem is stated with all its sub-components and then the system which this problem is emerged is described in this report. Making related assumptions, a step-by-step solving approach is undertaken with considering the managers approach to cleaning and purchasing. At each step, cleaning or buying one more additional tanker options are compared. In addition, considering business environment and its future, alternative actions are introduced. Considering all alternative actions, it could be stated that buying four D type tankers and increasing fleet size without any cleaning is recommended and this results with the total cost of $ 110.846,73. This is found to be the best alternative since it considers the effect of cleanings on workers health without comparing it with any financial gain. In addition considering the compartment sizes of D type tankers and currently having no D type tankers on hand, it is the recommended action for manager.

17

4. GLOSSARY
In this part, technical expressions which are used in the report will be explained. Mathematical model: Mathematical model is a description of a system using mathematical concepts and language. Decision variable: Decision variables are the variables within a mathematical model that one can control. They are not random variables and they are related to the managers decisions in this content. Constraint: Constraints are the conditions in an optimization environment which are needed to be satisfied. Infeasible: A mathematical model is said to be infeasible if it cannot satisfy all or some of the constraints it contains.

18

5. APPENDIX

PAGE Appendix A - Mathematical Model of the First Step Analysis ............................................ Appendix B - GAMS Output of the First Step ..................................................................... Appendix C - Mathematical Model Changes for the Second Step Analysis Appendix D - GAMS Output of the Second Step ................................................................ Appendix E - Mathematical Model Changes for the Third Step Analysis ........................... Appendix F - GAMS Output of the Third Step .................................................................... Appendix G - Mathematical Model Changes for the Fourth Step Analysis ........................ Appendix H - Mathematical Model Changes for the Fifth Step Analysis ........................... II IV V VI VII VIII IX IX

Appendix J - Mathematical Model Changes for the Six Step Analysis ................................ X Appendix K - GAMS Output of the Option 1 ...................................................................... XI

Appendix L - GAMS Output of the Option 2 ....................................................................... XII

Appendix - I

Appendix A - Mathematical Model of the First Step Analysis: Sets: i: r: j: Type of tankers {A, B, C, D} Routes defined in the question {1, 2, , 12} Depots {1, 2, , 11}

Parameters: Ni : CAi : CCi : Dr : DTr : Tj : Number of on hand tankers of type i Capacity of type i tankers for acidic Capacity of type i tankers for caustic Duration of route r Distance of route r Total demand of depot j

Decision Variables: Xir : Yir : Lr : Si : { Number of type i tankers assigned to route r Number of times a tanker assigned to route r completes its assigned route Number of type i tankers salvaged

Objective function: min Z = DTr * Lr * 0.1 + Yir * 200 - Si * 3000 Constraints: Yir Ni * Xir Yir Ni ( If type i is not assigned to route r, number of total tankers assigned to route r should be zero ) for all i (Sum of all type i tankers assigned to a route is smaller than its on hand amount)

Appendix - II

Yir * CCi * 5240 / Dr 53000 Yir * CCi * 5240 / Dr Tj

(Caustic demand of Tepe) for all j ( Acidic demand of all depots ) ( Maintenance restriction on total number of times a tanker assigned to route r completes the route) ( Number of unnecessary tankers of each type is salvaged ) ( Legal restriction on acidic )

Lr ( 5240 / Dr ) * Yir for all r

Si Ni - Yir for all i

16,5 for all j

16,5 for all j

( Legal restriction on caustic )

5,5 for all j

( Delivery amount of at most 5,5 for Hanya )

5,5 for all j for all j except Hanya and Mara3

( Delivery amount of at most 5,5 for Maras3 )

15

( Delivery amount of at least 15 tons for depots )

Xir , Yir , Lr , Si 0 THE ROUTES

Xir , Yir integer

1: {Tepe, Hanya} i.e: Tepe-Hanya-Tepe 2: {Tepe, Bor}

6: {Tepe, Hanya, Mara3} 7: {Tepe, Hanya, eme} 8: {Tepe, Hanya, Koru}

3: {Tepe, Bor, Hanya} i.e: Tepe-Bor-Hanya-Tepe 4: {Tepe, Hanya, Mara1} i.e: Tepe-Hanya-Mara1-Tepe OR Tepe-Mara1-Hanya-Tepe OR Tepe-Mara1-Tepe 5: {Tepe, Hanya, Mara2}

9: {Tepe,Lara} 10: {Tepe, Lara, Hanya} 11: {Tepe, Geyve} 12: {Tepe, Hora}

Appendix - III

Appendix B - GAMS Output of the First Step

GAMS Rev 227 x86/MS Windows 05/19/11 20:49:34 Page 5 G e n e r a l A l g e b r a i c M o d e l i n g S y s t e m Solution Report SOLVE system_1 Using MIP From line 162 S O L V E MODEL TYPE SOLVER system_1 MIP CPLEX S U M M A R Y OBJECTIVE DIRECTION FROM LINE z MINIMIZE 162

**** SOLVER STATUS **** MODEL STATUS **** OBJECTIVE VALUE

1 NORMAL COMPLETION 10 INTEGER INFEASIBLE 0.0000 0.140 0 1000.000 10000

RESOURCE USAGE, LIMIT ITERATION COUNT, LIMIT

ILOG CPLEX May 1, 2008 22.7.2 WIN 4792.4799 VIS x86/MS Windows Cplex 11.0.1, GAMS Link 34 Cplex licensed for 1 use of lp and barrier. Problem is integer infeasible. No solution returned

Appendix - IV

Appendix C - Mathematical Model Changes for the Second Step Analysis:

Additions
Decision Variable: Mi : New number of tankers of type i Ti : {

Ci : Number of type i tankers cleaned Costraints: Ci 10 * Ti MA = NA - CA + CB MB = NB + CA + CB MC = NC - CC + CD MD = ND - CD + CC Ci , Mi 0 ( If there is not any cleaning on type i, the number of cleaned tankers should be zero) ( Change in tanker numbers due to cleaning ) ( Change in tanker numbers due to cleaning ) ( Change in tanker numbers due to cleaning ) ( Change in tanker numbers due to cleaning ) Ti, Ci integer

Changes Made
Old: Si Ni - Yir for all i New: Si Mi - Yir for all i ( Number of unnecessary tankers of each type is salvaged ) ( Number of unnecessary tankers of each type is salvaged )

Appendix - V

Appendix D - GAMS Output of the Second Step

GAMS Rev 230 WIN-VIS 23.0.2 x86/MS Windows 05/21/11 17:41:38 Page 5 G e n e r a l A l g e b r a i c M o d e l i n g S y s t e m Solution Report SOLVE system_1 Using MIP From line 188 S O L V E MODEL TYPE SOLVER system_1 MIP CPLEX S U M M A R Y OBJECTIVE DIRECTION FROM LINE z MINIMIZE 188

**** SOLVER STATUS **** MODEL STATUS **** OBJECTIVE VALUE

1 NORMAL COMPLETION 10 INTEGER INFEASIBLE 0.0000 0.093 0 1000.000 10000

RESOURCE USAGE, LIMIT ITERATION COUNT, LIMIT ... Problem is integer infeasible. No solution returned

Appendix - VI

Appendix E - Mathematical Model Changes for the Third Step Analysis:

Additions
Decision Variable: Ji : Mi : Number of bought tankers of type i New number of tankers of type i

Costraints: M i = N i + Ji Ji , M i 0 for all i Ji integer ( Updating the total number of tankers )

Changes Made
Old: Objective function: min Z = DTr * Lr * 0.1 + Yir * 200 - Si * 3000

New: Objective function: min Z = DTr * Lr * 0.1 +( + Ji * 8000 Yir ]- Ji ) ) * 200 - Si * 3000

Appendix - VII

Appendix F - GAMS Output of the Third Step


S O L V E MODEL TYPE SOLVER system_1 MIP CPLEX S U M M A R Y OBJECTIVE DIRECTION FROM LINE 1 NORMAL COMPLETION 8 INTEGER SOLUTION 110846.7273 0.557 33 1000.000 10000 z MINIMIZE 174

**** SOLVER STATUS **** MODEL STATUS **** OBJECTIVE VALUE

RESOURCE USAGE, LIMIT ITERATION COUNT, LIMIT

MIP status(102): integer optimal, tolerance Fixed MIP status(1): optimal Solution satisfies tolerances. MIP Solution: Final Solve: Best possible: Absolute gap: Relative gap: 110846.727273 110846.727273 106318.938883 4527.788390 0.040847 (31 iterations, 0 nodes) (2 iterations)

---- VAR Y ... A.route8 ... B.route1 ... C.route3 ... C.route5 C.route6 ... C.route10 C.route11 C.route12 ... D.route3 D.route4 ... D.route7 ...

Number of type i tankers assigned to route r LOWER . . . . . . . . . . . LEVEL 1.000 3.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 2.000 UPPER 100.000 MARGINAL 3200.000

100.000 -3088.000 100.000 100.000 100.000 100.000 100.000 100.000 100.000 100.000 100.000 3200.000 3200.000 143.333 3200.000 3200.000 3200.000 3200.000 3200.000 3200.000

Appendix - VIII

Appendix G - Mathematical Model Changes for the Fourth Step Analysis:

Old: Objective function: min Z = DTr * Lr * 0.1 + Yir * 200 - Si * 3000 New: Objective function: min Z = DTr * Lr * 0.1 + Yir * 200 - Si * 3000 + 8000

On hand number of type D tankers is increased to one from zero.

Appendix H - Mathematical Model Changes for the Fifth Step Analysis:

Old: Objective function: min Z = DTr * Lr * 0.1 + Yir * 200 - Si * 3000 New: Objective function: min Z = DTr * Lr * 0.1 + Yir * 200 - Si * 3000 + 16000

On hand number of type D tankers is increased to two from one.

Appendix - IX

Appendix J - Mathematical Model Changes for the Six Step Analysis:

Old: Objective function: min Z = DTr * Lr * 0.1 + Yir * 200 - Si * 3000 New: Objective function: min Z = DTr * Lr * 0.1 + Yir * 200 - Si * 3000 + 24000

On hand number of type D tankers is increased to three from two.

Appendix - X

Appendix K - GAMS Output of the Option 1


GAMS Rev 230 WIN-VIS 23.0.2 x86/MS Windows 05/21/11 18:04:52 Page 5 G e n e r a l A l g e b r a i c M o d e l i n g S y s t e m Solution Report SOLVE system_1 Using MIP From line 188 S O L V E MODEL TYPE SOLVER system_1 MIP CPLEX S U M M A R Y OBJECTIVE DIRECTION FROM LINE 1 NORMAL COMPLETION 8 INTEGER SOLUTION 111646.7273 0.046 34 1000.000 10000 z MINIMIZE 188

**** SOLVER STATUS **** MODEL STATUS **** OBJECTIVE VALUE

RESOURCE USAGE, LIMIT ITERATION COUNT, LIMIT MIP Solution: Final Solve:

111646.727273 111646.727273

(32 iterations, 0 nodes) (2 iterations)

R 1 A B C D TANKERS 3 2 3 4 5

O 6

U 7

E 8 1

S 9 10 11 1 12 1

TOTAL TANKERS 3 3 4 4

1 1 1 1 1 1 2 Table Appendix K-1: Information gathered from the related GAMS Output

Appendix - XI

Appendix L - GAMS Output of the Option 2

GAMS Rev 230 WIN-VIS 23.0.2 x86/MS Windows 05/21/11 18:18:02 Page 5 G e n e r a l A l g e b r a i c M o d e l i n g S y s t e m Solution Report SOLVE system_1 Using MIP From line 188 S O L V E MODEL TYPE SOLVER system_1 MIP CPLEX S U M M A R Y OBJECTIVE DIRECTION FROM LINE 1 NORMAL COMPLETION 8 INTEGER SOLUTION 111646.7273 0.062 35 1000.000 10000 z MINIMIZE 188

**** SOLVER STATUS **** MODEL STATUS **** OBJECTIVE VALUE

RESOURCE USAGE, LIMIT ITERATION COUNT, LIMIT MIP Solution: Final Solve: Best possible: Absolute gap: Relative gap:

111646.727273 111646.727273 110681.488289 965.238984 0.008645

(33 iterations, 0 nodes) (2 iterations)

R 1 A B C D TANKERS 3 2 3 4 5

O 6

U 7

E 8

S 9 10 11 1 12 1

TOTAL TANKERS 2 3 4 5

1 1 1 1 1 1 2 1 Table Appendix L-1: Information gathered from the related GAMS Output

Appendix - XII

You might also like