Abstract
We study scheduling and due-window assignment problems with the objective function of a minmax type, i.e., the goal is to minimize the largest cost among all scheduled jobs. We assume that the processing times of jobs are position-dependent in the most general way. For a single machine and for a proportionate flow shop environment we present polynomial time solution procedures that are based on solving a linear assignment problem as a subroutine. We further extend the single machine model by allowing job-rejection, provided that the processing times deteriorate, i.e., the position-dependent processing times are non-decreasing functions of the job position. For this setting, the scheduler may decide not to process certain jobs, and each rejected job is penalized accordingly. The problem with the objective that additionally involves a maximum rejection cost component is also shown to be solvable in polynomial time.
Similar content being viewed by others
References
Biskup D (1999) Single machine scheduling with learning considerations. Eur J Oper Res 115:173–178
Cao Z, Zhang Y (2007) Scheduling with rejection and nonidentical job arrivals. J Syst Sci Complex 20:529–535
Cao Z, Wang Z, Zhang Y, Liu S (2006) On several scheduling problems with rejection or discretely compressible processing times. Lect Notes Comput Sci 3959:90–98
Cheng Y, Sun S (2009) Scheduling linear deteriorating jobs with rejection on a single machine. Eur J Oper Res 194:18–27
Cheng TCE, Ding Q, Lin BMT (2004) A concise survey of scheduling with time-dependent processing times. Eur J Oper Res 152:1–13
Gabow HN, Tarjan RE (1988) Algorithms for two bottleneck optimization problems. J Algorithms 11:411–417
Gawiejnowicz S (1996) A note on a single processor with speed dependent on a number of executed jobs. Inf Process Lett 57:297–300
He C, Leung Y-T, Lee K, Pinedo ML (2016) Improved algorithms for single machine scheduling with release dates and rejection. 4OR 14:41–55
Janiak A, Marek M (2004) Parallel processor scheduling problems with optimal due interval assignment. In: Proceedings of the 5th Mexican international conference in computer science (ENC’04). IEEE Press, Colima, pp 383–387
Janiak A, Kovalyov MY, Marek M (2007) Soft due window assignment and scheduling on parallel machines. IEEE Trans Syst Man Cybern A Syst Hum 37:614–620
Janiak A, Janiak WA, Kovalyov MY, Werner F (2012) Soft due window assignment and scheduling of unit-time jobs on parallel machines. 4OR Q J Oper Res 10:347–360
Janiak A, Janiak WA, Krysiak T, Kwiatkowski T (2015) A survey on scheduling problems with due windows. Eur J Oper Res 242:347–357
Khuller S, Mestre J (2008) An optimal incremental algorithm for minimizing lateness with rejection. Lect Notes Comput Sci 5193:601–610
Lu L, Zhang L, Yuan J (2008) The unbounded parallel batch machine scheduling with release dates and rejection to minimize makespan. Theoret Comput Sci 396:283–289
Lu L, Cheng TCE, Yuan J, Zhang L (2009) Bounded single machine parallel-batch scheduling with release dates and rejection. Comput Oper Res 36:2748–2751
Mosheiov G (2001a) A common due-date assignment problem on parallel identical machines. Comput Oper Res 28:719–732
Mosheiov G (2001b) A due-window determination in minmax scheduling problems. INFOR 39:107–123
Mosheiov G (2011) Proportionate flow shops with general position-dependent processing times. Inf Process Lett 17:174–177
Mosheiov G, Sarig A (2009) Minmax scheduling problems with a common due-window. Comput Oper Res 36:1886–1892
Mosheiov G, Sarig A (2010) Scheduling with a common due-window: polynomially solvable cases. Inf Sci 180:1492–1505
Ou J, Zhong X, Wang G (2015) An improved heuristic for parallel machine scheduling with rejection. Eur J Oper Res 241:653–661
Panwalkar SS, Dubek RA, Smith ML (1973) Sequencing research and the industrial scheduling problem. In: Elmaghraby SE (ed) Symposium on the theory of scheduling and its applications. Springer, Berlin, pp 29–37
Panwalkar SS, Smith ML, Koulamas C (2013) Review of the ordered and proportionate flow shop scheduling research. Nav Res Logist 60:46–55
Pinedo ML (2017) Scheduling: theory, algorithms and systems. Prentice-Hall, Englewood Cliffs
Sengupta S (2003) Algorithms and approximation schemes for minimum lateness/tardiness scheduling with rejection. Lect Notes Comput Sci 2748:79–90
Shabtay D (2014) The single machine serial batch scheduling problem with rejection to minimize total completion time and total rejection cost. Eur J Oper Res 233:653–661
Shabtay D, Gaspar N (2012) Two machine flow shop scheduling with rejection. Comput Oper Res 39:64–74
Shabtay D, Gaspar N, Kaspi M (2013) A survey on offline scheduling with rejection. J Sched 16:3–28
Yin Y, Xu D, Sun K, Li H (2009) Some scheduling problems with general position-dependent and time-dependent learning effects. Inf Sci 179:2416–2425
Zhang L, Lu L (2016) Parallel-machine scheduling with release dates and rejection. 4OR 14:165–172
Zhang L, Lu L, Yuan J (2009) Single machine scheduling with release dates and rejection. Eur J Oper Res 198:975–978
Zhang L, Lu L, Yuan J (2010) Single-machine scheduling under the job rejection constraint. Theoret Comput Sci 411:1877–1882
Acknowledgements
This research was supported in part by the Recanati Fund of The School of Business Administration, and The Charles I. Rosen Chair of Management, The Hebrew University, Jerusalem, Israel.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix A: Numerical examples
Appendix A: Numerical examples
Example 1
Consider a 7-job problem. The position-dependent processing times are given in Table 2. Assume: \( \alpha = 3, \beta = 1.2, \gamma = 2, \delta = 1.5 \). Following Algorithm \( Due\_Window\_p_{jr} \), Case (i) is optimal. It follows that \( d_{1} = d_{2} = 0, \) the optimal job sequence is \( \left( {7,2,1,5,6,4,3} \right) \), \( C_{max} = 32 \), and \( Z = \beta C_{max} = 38.4 \).
Example 2
The job-position processing times are those of Example 1. Assume: \( \alpha = 2, \beta = 3.2, \gamma = 0.6, \delta = 1.8 \). Now Case (ii) is optimal. After solving seven assignment problems (\( ASSIGN\left( k \right), k = 1, \ldots ,7 \)), we obtain the following optimal solution (achieved for \( k = 2 \)): \( d_{1} = d_{2} = 33.69, \) the optimal job sequence is \( \left( {2,7,1,5,6,4,3} \right) \), \( C_{ \text{min} } = 22, C_{ \text{max} } = 41 \), and \( Z = \frac{{\beta \left( {\alpha + \gamma } \right)}}{\alpha + \beta }C_{ \text{max} } - \frac{{\alpha \left( {\beta - \gamma } \right)}}{a + \beta }C_{ \text{min} } = 43.6 \).
Example 3
The job-position processing times are those of Example 1, and \( \alpha = 2, \beta = 3, \gamma = 1.5, \delta = 2 \). Case (iii) is optimal. As in Example 2, we solve seven assignment problems. The optimal solution is obtained for \( k = 7 \): \( d_{1} = C_{ \text{min} } = 6; d_{2} = C_{ \text{max} } = 32, \) the optimal job sequence is \( \left( {7,2,1,5,6,4,3} \right) \), and \( Z = \delta C_{ \text{max} } - \left( {\delta - \gamma } \right)C_{ \text{min} } = 61 \).
Example 4
The job-position processing times are those of Example 1. Assume: \( \alpha = 3, \beta = 2, \gamma = 2.5, \delta = 1 \). Case (iv) is optimal. As in Example 1, we solve a single assignment problem in order to minimize the makespan. The optimal solution is: \( d_{1} = 0; d_{2} = C_{ \text{max} } = 32, \) the optimal job sequence is \( \left( {7,2,1,5,6,4,3} \right) \), and \( Z = \delta C_{ \text{max} } = 32 \).
Example 5
Consider a 5-machine proportionate flow shop 4-job problem. The position-dependent processing times are given in Table 3. The cost parameters are those given in Example 2: \( \alpha = 2, \beta = 3.2, \gamma = 0.6, \delta = 1.8 \). Again, Case (ii) is optimal, and thus \( d_{1} = d_{2} = \frac{{\alpha C_{ \text{min} } + \beta C_{ \text{max} } }}{\alpha + \beta } \), and \( Z = \frac{{\beta \left( {\alpha + \gamma } \right)}}{\alpha + \beta }C_{ \text{max} } - \frac{{\alpha \left( {\beta - \gamma } \right)}}{a + \beta }C_{ \text{min} } \). We first solve the bottleneck problem, and obtain an optimal value of 7. Hence, only the job-positions with processing time of 7 and above should be considered as potential pseudo job-positions. Following Algorithm \( Due\_Window\_p_{jr} \_Flowshop \), we obtain: Job 3 is scheduled first; the pseudo job-position is (4, 4) (its value is \( 5 \times 7 = 35) \); the optimal job sequence is (3, 2, 1, 4); \( C_{max} = 46 \), \( C_{min} = 30 \) (obtained by delaying the operations on machine 5 to eliminates idle times), and \( Z = 43.6 \).
Example 6
Consider a 7-job problem. The (deteriorating) job-position processing times are given in Table 4. The rejection costs (after sorting) are: \( e_{1} = 42, e_{2} = 59, e_{3} = 67, e_{4} = 78, e_{5} = 82, e_{6} = 90, e_{7} = 95 \). Assume that \( \alpha = 3, \beta = 1.2, \gamma = 2, \delta = 1.5 \). Given these cost parameters, Case (i) is optimal. The initial values for the binary search are: \( k_{ \text{min} } = 0 \) (no jobs are rejected), \( k_{ \text{max} } = 7 \) (all jobs are rejected). Thus, \( k = \frac{{k_{ \text{min} } + k_{ \text{max} } }}{2} = 4 \), and we solve the due-window assignment problem for jobs \( \left\{ {5,6,7} \right\} \). The resulting makespan is \( C_{ \text{max} } = 34 \). Since \( Z^{\left( 4 \right)} = \beta C_{ \text{max} } = 40.8 < 78 = e_{4} \), we check the option of rejecting less jobs. In the next iteration \( k_{ \text{min} } = 0 \) and \( k_{ \text{max} } = 4 \), implying \( k = 2 \). We solve the due-window assignment problem for jobs \( \left\{ {3,4,5,6,7} \right\} \). The resulting makespan is \( C_{ \text{max} } = 84 \). In this case: \( Z^{\left( 2 \right)} = \beta C_{ \text{max} } = 100.8 > 59 = e_{2} \), and we check the option of rejecting more jobs. Now \( k_{ \text{min} } = 2 \) and \( k_{ \text{max} } = 4 \), implying \( k = 3 \). We solve the due-window assignment problem for jobs \( \left\{ {4,5,6,7} \right\} \). The resulting makespan is \( C_{ \text{max} } = 55 \), and \( Z^{\left( 3 \right)} = \beta C_{ \text{max} } = 66 < 67 = e_{3} \). We check again the option of rejecting less jobs: \( k_{ \text{min} } = 2 \) and \( k_{ \text{max} } = 3 \). Thus, the next \( k \)-value to be checked is 3, which is equal to \( k_{ \text{max} } \) and the procedure is finished. The optimal solution consists of rejecting jobs \( \left\{ {1,2,3} \right\} \), and processing the remaining jobs: \( \left\{ {4,5,6,7} \right\} \). The due-window is reduced to a due-date at time zero (\( d_{1} = d_{2} = 0 \)), the optimal job sequence is \( \left\{ {4,7,5,6} \right\} \) with maximum (tardiness) cost of 66. Thus, the optimal minmax cost is 67, obtained by the rejection cost of job 3.
Rights and permissions
About this article
Cite this article
Mosheiov, G., Sarig, A. & Strusevich, V. Minmax scheduling and due-window assignment with position-dependent processing times and job rejection. 4OR-Q J Oper Res 18, 439–456 (2020). https://doi.org/10.1007/s10288-019-00418-w
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10288-019-00418-w
Keywords
- Scheduling
- Sequencing
- Due-window assignment
- Position-dependent processing times
- Proportionate flow shop
- Job-rejection