APPLICATION

Akyama? Engineering & Construction Firm in the construction sector in Aksaray, Turkey is planning on purchasing heat and noise insulation materials for a new project. Firm management is desiring to make a new buying plan based on three criteria from five different suppliers that it worked in the past period. The business top manager wants to be under-affected by delays during product delivery due to Covid-l9. Therefore, the delivery time of the products in the last 8 months has been examined, and ‘On-Time Delivery Performance Level of the Product (%)’ has been determined as the first criteria for each supplier. In order to reduce the negative effects of the human and environmental health of the material to be purchased, the criteria of ‘Green Product Quality' and ‘Occupational Safety and Occupational Health Practices’ applied by suppliers due to the epidemic have been scored as *%’. Information about criteria and unit price of the material to be purchased from suppliers are given in Table 7.2.

The business has detected that the demand for the products will be 450 at least and has budgeted $26,000 in total for the material to be purchased from the suppliers. Besides, according to previous period information of the business, as per agreements with suppliers, at least 100 from Supplier-1, at least 130 from Supplier-2, and at most 170 from Supplier-5 are purchased. A minimum of 140 and a maximum of 180 from Supplier-3, a minimum of 155, and a maximum of 165 from Supplier-4 can be purchased. The number of materials to be purchased from each supplier is given in Table 7.3.

Firm management is planning on increasing the amount to be purchased from Supplier-3, Supplier-4 and Supplier-5 according to the buying plan. The tolerance values for these suppliers are 15, 20, 30, respectively. The business has allocated an

TABLE 7.2

Performance Information of Suppliers

TABLE 7.3

The Number of Materials to be Purchased from Suppliers

additional budget, which is $4000. According to the information given above, the Multiobjective Sustainable Supplier Selection problem (S-SSP) of the business is created as follows:

Subject to (PI)

Step 1: In this step, the positive ideal solution set is determined. Solutions determined for each goal function in (PI) are given below.

The ideal solution set of (PI) is obtained as Z^l* = {51,895; 52,410; 57,905}. Step 2: In this step, compromised solutions are determined by depending on Z¹* = {51,895; 52,410; 57,905}. Namely, the solution is made forGCM (for p = 1) and Compromise Programming (CP) (forp = »). While determining the longest distance for p =1, the longest distance is completely dominated for p = °o.

i. forp = 7; (PI) is organized as follows.

In order global objective function to be minimum, (0.004256х, + 0.003969x₂+0.004086x₃+0.004269x.i+0.004602x₅) should be maximum. According to this, the model below is obtained.

Subject to (PI.la)

The variable values obtained from the solution of (PI.la) are specified as X| = 102; x₂ = 160; x₃ 140; x₄ = 155; x₅ = 146; Max F = 2.99771. Accordig to these variable values, x" =(102; 160; 140; 155; 146) ve Z¹¹ =(51,775; 52,410; 7,905) are obtained. Moreover, Min G: [3-(2.997717 )J is 0.002283. Accordingly, the goal functions are very close to their positive ideal solutions.

ii. forp = °o; (PI) is organized as follows.

Subject to (PI.lb)

The variable values and objective function values specified from the solution of (PI.lb) are obtained as = (114; 146; 140; 155; 14 and Z'⁰⁰ = (51,835; 52,320; 57,865). Besides. Min d„ = 0.001717 shows that objective functions are close to their own positive ideal solutions.

Step 3: Similar to Step 1, find the PIS under the constraints w'ith maximum tolerances. Maximum amounts to be purchased from Supplier-3, Supplier-4 and Supplier-5 are demanded to be increased as 15, 20 and 30, respectively. Due to these increases, the budget has been increased $4000. According to these information, (PI) is organized as follows.

Subject to (P2)

In (P2), solutions made for each objective function are given below.

Depending on these solutions, PIS set of (P2) are given below.

Step 4: In this step, by using Z°* ={60,475; 62,395; 68,565}, Global Criterion method and Compromise Proramming solutions are determined.

i. for p = 1, (P2) is organized as follows.

Subject to (P2.1a)

The variable values obtained from the solution of (P2.la) are determined as x, = 104; x₂ = 291; x₃ = 140; x₄ = 155; x₅ = 145; MaxF = 2.988625. According to these variable values, Min r/_: = [3-(2.988625 )J = 0.011375. Based on this information, x⁰¹ =(104; 291; 140; 155; 145) and Z⁰¹ = (59,785; 62.395; 68.565) are obtained.

ii. For p = (P2) is organized as follows.

Subject to (P2.1b)

The variable values and objective function values determined from the solution of (P2.1b) are дг°~ =(158; 228; 140; 155; 145) and Z^0o° = (60,055; 61,990; 68,385). Min d„ = 0.006945 is obtained.

Step 5: In this step, the fuzzy linear programming solution is made for each objective function. At this stage; firstly, ideal solution sets obtained from Steps 1 and 3 are used. These are Z¹* ={51,895; 52,410; 57,905} and Z°*= {60,475; 62,395; 68,565}. Membership functions are defined according to positive ideal solutions (Figures 7.3-7.8).

Membership function for Z,(x);

Membership function for Z₂(a);

FIGURE 7.2 Fuzzy membership function for Z, (.v).

FIGURE 7.3 Fuzzy membership function forZ₂(x).

FIGURE 7.4 Fuzzy membership function for Z,(x)-

FIGURE 7.5 Membership function for budget constraint.

FIGURE 7.6 Fuzzy membership function for constraint g,,(x)-

FIGURE 7.8 Fuzzy membership function for constraint gio(*)- Membership function for Z₃(;r);

Membership functions are then created for constraint functions. Membership function for budget constraint;

Membership function for g₆(x);

Membership function for g_s(x);

Membership function for g₁₀ (x);

Using fuzzy membership functions, the problem is arranged mathematically as follows.

Subject to (P3)

The results obtained from the Fuzzy Linear Programming solution for each objective function (P3) are given below'.

For Max Z,(.r);

For Max Z₂ (.v);

For Max Z, (a);

Based on these results; a' = max(0.604895;0.5;0.5) = 0.604895 is chosen. Step 6: Final compromise solution steps for a* = 0.604895 are given below. Firstly (PI), if the membership functions in step 5 are arranged fora* -cut = 0.604895. the mathematical model named (P4) is obtained.

Subject to (P4)

Positive ideal solutions for (P4) are determined at first.

Max Z (a) igin;

Max Z₂ (a) igin;

Max Zi(a) igin;

Based on these results, positive ideal solution set of (P4) is (54925; 56375; 62125). By using positive ideal solution set;

i. For p = 1;

Subject to (4.1a)

From the solution of (P4.1a), X = 100; x₂ = 215; дг₃ = 140; x₄ = 155 ve x₅ = 145 variable values are obtained. According to this, each objective function is obtained as Z = 54,925; Z =56,375; Z = 62,125. Min G.(3-3) = 0 indicates that all objectives realize in their own positive ideal solution.

ii. For /) = “

Subject to (4.1b)

The variable values obtained from the solution of (4.1b) are x, = 100; x₂ = 215; x₃ = 140; x₄ = 155; x₅ = 145. On the other hand, each objective function value is obtained as Z, = 54,925;Z₂ = 56,375; Z₃ = 62,125. Each objective function value of (P4.1b) is designated as = 0. Thereby, each objective has realized in its own positive ideal solution.

With the Fuzzy GCM algorithm consisting of six steps, variable values and goal function values obtained in each step are given in Table 7.4.

It is possible to analyze the results in Table 7.4 by dividing them into three groups.

i. In Step-1, a positive ideal solution set for (PI) has been designated. These values indicate the best performance of each objective function. Due to the fact that objective function values realize in different values of the variables, there is no feasible solution in this step. Actually, non- dominated solutions can be determined using these information. By using the information in Step-1, non-dominated solutions table is given in Table 7.5.

The values in Table 7.5 include the results that Global Criterion Method and CP will give in terms of nondominated solution. This situation is

TABLE 7.4

Fuzzy Global Criterion Method Results

TABLE 7.5

Non-Dominated Solution of (PI)

explainable by using the information in Steps-1 and 2. Firstly, (%) value of each objective function has been calculated based on the results in Step-1.

These rates show the average % value of the purchased material. In other words, they show the average performance value of each objective. The calculations can be made for Table 7.5 (%) transformation of non-dominated solutions are given in Table 7.6.

Ranges of each objective function from the results in Table 7.6 that realizes in terms of Global Criterion Method and CP are determined as 73.75356 2 < 74.82808; 82.48575 2 <82.84384.

From the information from Step-2,

for p = 1 (Global Criterion Method),

TABLE 7.6

Nondominated Solutions in Terms of Percentage

These results have realized in lower limits of % ranges of non- dominated solutions.

For p = ~ (CP).

The % values obtained in terms of CP has been within the % ranges of the non-dominated solutions and higher than the lower limits. Two solution results are an expected situation in terms of distance family. The solution obtained for p = ~ is dominate the results given by P= 1-

ii. The difference in these steps stems from increasing right hand side values of gi(.v), gb(x), gs(x) and gio(J0 constraints in accordance with the information coming from the decision maker. The values designated in these steps are given in Table 7.4. It is possible to make similar calculations and comments given above for Steps 3 and 4.

iii. In Steps 5 and 6, the decision is made according to the positive ideal solutions obtained from Steps 1 and 3. From the fuzzy linear programming perspective in Step-5, firstly, membership functions were created for each goal function using Z^|S ={51895; 52410; 57905} and Z°* ={60,475; 62,395; 68,565} afterwards, fuzzy model was established by defining membership functions for constraint functions. In Step-5, the established model is solved for each objective function, and the most satisfactory one is preferred among these solutions. In Step-6, firstly, by using cc -cut determined from a = max*a{, right hand side values of giU), ge(x), gg(x) and gioOO constraint functions have been assigned.

In (P4), each goal function is realized at the same value of same decision variables. According to Lai and Hwang (1994), this solution type in MODM problems are called optimal solution. Therefore, the solutions obtained from (P4.la) and (P4.1b) are optimal solutions. When these results are taken into consideration, it is seen that positive ideal solution set and compromised solutions are equal to each other in the problem. Objective function values for these three situations have been calculated below as %.

Accordingly, when 100, 215, 140, 155 and 145 material are purchased respectively, on-time delivery performance mean of products are calculated as 72.74834%, green product quality mean is 74.66887%, and Work Safety and Labor Health average performance of the suppliers are 82.48575%.