Sei sulla pagina 1di 5

1. For this problem, Total processing time = 179 < 240 So there will be no tardiness for any job.

Then LPT will give us the optimal: {J21 , J14 , J22 , J11 , J18 , J20 , J19 , J4 , J1 , J8 , J13 , J17 , J2 , J10 , J12 , J3 , J15 , J6 , J9 , J5 , J7 , J16 }

2.

First put all the job on the machine that can process them the fastest. Then we get:

Then for Machine 2, choose the job that has the longest processing time on one of M1,M3 that has the shorter processing time with the job. Then we get:

Repeat this process:

Then choose the job that has the longest processing time on M2, putting it on M3.

Next, we alternatively choose jobs between M1 and M2, putting them on M3. When the completion time for M2 is larger than M1, then we choose job with largest processing time on M2, vice versa. Then we get:

This is a good scheduling: M1: {J10 J5 J4} Completion time = 31 M2: {J1 J2 J3 J6 J7 J12} Completion time = 34 M3: {J8 J13 J9 J11} Completion time = 33

So the completion time for all the jobs that can be done is 34.

3. Use LPT to solve the problem. There are m machines and 2m+1 jobs. The shortest processing time is m and the largest processing time is 2m 1 (J1 and J2 ). Then we assign the srt m jobs that has the larest processing time on m machines: If m is odd, then for the rst 2m jobs: Machine Job (rst) Processing time Job (second) Processing time M1 J1 2m-1 J2m 2m 1 + m = 3m 1 M2 J2 2m-1 J2m1 3m-1 M3 J3 2m-2 J2m2 3m-1 M4 J4 2m-2 J2m3 3m-1 . . . . . . . . . . . . . . . 3m1 +1 Mm1 J m1 Jm+2 3m-1 2 3m1 3m1 Mm Jm Jm+1 + 3m1 = 3m 1 2 2 2

If m is even, then for the rst 2m jobs: Machine Job (rst) Processing time M1 J1 2m-1 M2 J2 2m-1 M3 J3 2m-2 M4 J4 2m-2 . . . . . . . . . 3m Mm1 J m1 2 3m Mm Jm 2 2

Job (second) J2m J2m1 J2m2 J2m3 . . . Jm+2 Jm+1

Processing time 2m 1 + m = 3m 1 3m-1 3m-1 3m-1 . . . 3m 3m +( 2 1) = 3m 1 2 3m 3m 2 +( 2 1) = 3m 1

Hence, when we use LPT for the rst 2m jobs, each of the m machines has the same the completion time 3m 1. Then there is the J2m+1 that has not been asigned. Assgn J2m+1 to any of the m machines. Finally the completion time will be (3m 1) + processing time f or J2m+1 = (3m 1) + m = 4m 1 . By now, we have shown that the longest processing time schedule has Cmax = 4m1.

Next, the total processing time for all the 2m+1 machines is [(2m1)+m]m 2 2 + m = 3m2 By using the list (J1 , J2 , . . . .Jm1 , J2m1 , J2m , Jm , Jm+1 , . . . ., J2m2 , J2m+1 ), 2 this scheduling reaches the best possible Cmax = 3m = 3m m Hence, 4m 1 4 1 RA(I) = = 3m 3 3m is the best result for the longest processing time algorithm.

4. Suppose that there are 15 jobs denoted by xi , (i = 1, 2, ...15). Use Excel to do the simulation. Use Round(norminv(rand(), 2, 0.2), 2) to represent the processing time for machine 1 and use Round(norminv(rand(), 2, 0.3), 2) for machine 2. Compare buer size 0 and buer size 1, Trial Buer size 0 Buer size 1 Buer size 2 or more 1 34.86 33.18 33.18 2 35.49 34.22 34.22 3 34.19 32.59 32.59 4 34.17 33.1 33.1 5 34.5 32.64 32.64 6 33.36 31.84 31.84 7 34.94 33.48 33.48 8 36.04 34.27 34.27 9 34.28 33.44 33.44 10 34.8 33.69 33.69 Average 34.663 33.445 33.445

With buer size capacity of 1 job, Cmax will be smaller than with buer size 0. And for 15 jobs, buer size 1 is enough.

If we can interchange the machines: When the buer size is 0,

Trial Machine order in : M 1 M 2 1 33.98 2 34.93 3 32.71 4 33.48 5 34.01 6 33.56 7 33.22 8 35.03 9 35.47 10 34.86 Average 34.125 10th trial in excel:

Machine order in : M 2 M 1 34.09 34.88 33.32 33.08 34.13 34.6 34.04 34.98 35.02 34.61 34.275

When the buer size is 1,

We do the simulation 10 times, here is the results in excel:

Trial 1 2 3 4 5 6 7 8 9 10 Average

Machine order in : M 1 M 2 32.46 32.07 36.11 32.4 31.89 32.55 32.25 31.77 32.54 32.67 32.671

Machine order in : M 2 M 1 31.62 32.73 35.47 33.22 31.94 32.03 31.87 32.31 33 33.03 32.722

10th trial in excel:

Hence it is better in the order M 1 M 2 for the machine.

5. Use LPT algorithm:

Potrebbero piacerti anche