Академический Документы
Профессиональный Документы
Культура Документы
1/ 17
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
If the service a job receives is incomplete or unsatisfactory, the job feeds back
service node arrival
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
queue
server
departure
feedback
2/ 17
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
Model Considerations
When feedback occurs the job joins the queue consistent with the queue discipline The decision to depart or feed back is random with feedback probability
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3/ 17
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
Model Considerations
Feedback is independent of past history In theory, a job may feed back arbitrarily many times Typically is close to 0.0 GetFeedback Method
int GetFeedback(double beta) /* 0.0 <= beta < 1.0 */ { SelectStream(2); if (Random() < beta) return (1); /* feedback occurs */ else return (0); /* no feedback */ }
4/ 17
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
Statistical Considerations
Index i = 1, 2, 3, . . . counts jobs that enter the service node
fed-back jobs are not recounted
Jobs from outside the system are merged with jobs from the feedback process The steady-state request-for-service rate is larger than by the positive additive factor x Note that s increases with feedback but 1/ is the average service time per request
5/ 17
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
Example 3.3.1
job index arrival/ feedback service completion 1 1 9 10 2 3 3 13 3 4 2 15 4 7 4 19 5 10 7 26 13 5 31 6 14 6 37 15 3 40 7 19 4 44 8 24 6 50 26 3 53 9 30 7 60
6/ 17
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
Example 3.3.2
Program ssq2 was modied to incorporate immediate feedback
Interarrivals = Exponential(2.0) Service times = Uniform(1.0, 2.0)
1.0 0.9 0.8 x 0.7 0.6 0.5 0.00 0.05 0.10 0.15 0.20 0.25
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.0 8.0 6.0 q 4.0 2.0 0.0 0.00 0.05 0.10 0.15 0.20 0.25
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . .
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
Jobs ow into the service node at the average rate of To remain ow balanced jobs must ow out of the service node at the same average rate The average rate at which jobs ow out of the service node is x (1 ) Flow balance is achieved when = x (1 ) Saturation occurs when x = 1 or as 1 / = 0.25
8/ 17
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
9/ 17
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ........... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10
12
The last order is assumed to have no lag We assume that orders are delivered before the next inventory review With this assumption, there is no change to the specication model
Discrete-Event Simulation: A First Course Section 3.3: Discrete-Event Simulation Examples 10/ 17
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
Statistical Considerations
li remain correct If li 1 s the equations for li+ and When delivery lag occurs the time-averaged holding and shortage intervals must be modied
The delivery lag for interval i is 0 < i < 1
S
. . . . . . . . . . . . . . . . . . i i . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
l(t)
s
. . . . . . . . . li1 . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . li1 i di . . . . . . . . . . . . . . .
S d S di
. . . .
| lag, i | i1 i 1 + i
t i
11/ 17
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
Consistency Checks
Using the library rngs facilitates this kind of comparison It is a good practice to check for intuitive small-perturbation consistency
Use a small, but non-zero and check that appropriate statistics are slightly larger
12/ 17
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
Example 3.3.3
For the SIS with delivery lag, i = 0.0 i no order during i th interval, 0 < i < 1.0 otherwise The SIS is lag-free i i = 0.0 for all i If (S , s ) are xed then, even with small delivery lags:
, and u o , d are the same regardless of delivery lag Compared to the lag-free system, l + will decrease Compared to the lag-free system, l will increase or remain unchanged
13/ 17
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
Example 3.3.4
Delivery lags are independent Uniform(0.0, 1.0) random variates
2800 2600 2400 dependent cost, $ 2200 2000 1800 1600 0 5
no delivery lag
S = 80
10 15 20 25 30 35 40 45 50 Inventory parameter, s
55 60
Delivery lag causes l + to decrease and l to increase or remain the same Chold = $25 and Cshort = $700 cause shift up and to the left
Discrete-Event Simulation: A First Course Section 3.3: Discrete-Event Simulation Examples 14/ 17
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
queue
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
server
Assume repair times are Uniform(1.0, 2.0) random variates There are M machines that fail after an Exponential(100.0) random variate
Discrete-Event Simulation: A First Course Section 3.3: Discrete-Event Simulation Examples 15/ 17
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
Program ssms
Program ssms simulates a single-server machine shop The library rngs is used to uncouple the random processes The failure process is dened by the array failures
A O(M ) search is used to nd the next failure Alternate data structures can be used to increase computational eciency
16/ 17
SSQ with Immediate Feedback A simple inventory system with delivery lag A single-server machine shop
Example 3.3.5
The time-averaged number of working machines is M l
80 70 60 50 M l 40 30 20 10 0
. .. . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. ..... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . .
For small values of M the time-averaged number of operational machines is essentially M For large values of M this value is essentially constant at approximately 67
17/ 17