Nurse Rostering Benchmark Instances

Instances 1..24

These instances are available in a plain text file format as well as an XML format. See also :

  • - A zip file of all the instances.
  • - The file format guide and how to interpret the constraints and objectives.
  • - How to verify new solutions.
  • - An initial set of results and an integer programming model for these instances is here (solutions).
  • - New results are announced on the changes page.
Instance Weeks Employees Shift types Best known lower bound Best known solution
Instance1 txt xml 2 8 1 607 607
Instance2 txt xml 2 14 2 828 828
Instance3 txt xml 2 20 3 1001 1001
Instance4 txt xml 4 10 2 1716 1716
Instance5 txt xml 4 16 2 1143 1143
Instance6 txt xml 4 18 3 1950 1950
Instance7 txt xml 4 20 3 1056 1056
Instance8 txt xml 4 30 4 1300 1300
Instance9 txt xml 4 36 4 439 439
Instance10 txt xml 4 40 5 4631 4631
Instance11 txt xml 4 50 6 3443 3443
Instance12 txt xml 4 60 10 4040 4040
Instance13 txt xml 4 120 18 1348 1348
Instance14 txt xml 6 32 4 1278 1278
Instance15 txt xml 6 45 6 3827 3828
Instance16 txt xml 8 20 3 3225 3225
Instance17 txt xml 8 32 4 5746 5746
Instance18 txt xml 12 22 3 4459 4459
Instance19 txt xml 12 40 5 3149 3149
Instance20 txt xml 26 50 6 4769 4769
Instance21 txt xml 26 100 8 21133 21133
Instance22 txt xml 52 50 10 30241 30241
Instance23 txt xml 52 100 16 16990 17428
Instance24 txt xml 52 150 32 26571 42463

All solutions are verified before being reported on this website. It is not possible to verify lower bounds.

References

  • Strandmark, P., Qu, Y. and Curtois, T. First-order linear programming in a column generation-based heuristic approach to the nurse rostering problem. Computers & Operations Research, 2020. 120, p. 104945. (pdf)
  • Demirović, E., Musliu, N., and Winter, F. Modeling and solving staff scheduling with partial weighted maxSAT. Annals of Operations Research, 2019. 275(1): p. 79-99.
  • Smet P. Constraint reformulation for nurse rostering problems, in: PATAT 2018 twelfth international conference on the practice and theory of automated timetabling, Vienna, August, 2018, p. 69-80.
  • Rahimian, E., Akartunalı, K., and Levine, J. A hybrid integer programming and variable neighbourhood search algorithm to solve nurse rostering problems. European Journal of Operational Research, 2017. 258(2): p. 411-423.

 

Other instances

Best known solutions
GPost
File GPost.ros
Problem GPost
Employees 8
Schedule length 4 weeks
Cover type Cover is specified per shift.
5 html xml
7 html xml
8 html xml
Other versions GPost-B.ros Same as GPost.ros but without the requests on the first two days.
3 html xml
5 html xml
SINTEF
File SINTEF.ros
Problem SINTEF
Employees 24
Schedule length 3 weeks
Cover type Cover is specified per shift.
0 html xml
1 html xml
2 html xml
ORTEC
File ORTEC01.ros
Problem ORTEC
Employees 16
Shift types 4
Period 31 days
Cover type Cover is specified per shift.
References [BUR06, BUR07, BUR07b, BUR07c, FIJ06]
270 html xml
280 html xml
285 html xml
290 html xml
Other versions ORTEC02.ros Identical to ORTEC01 except employee B has five "Vacation" shifts in the first full week.
270 html xml
290 html xml
QMC
File QMC-1.ros
Problem QMC
Employees 19
Shift types 3
Period 28 days
Cover type Cover is specified per shift.
13 html xml
14 html xml
15 html xml
16 html xml
Other versions QMC-2.ros An alternative formulation with a couple of different constraints and in which cover may deviate from a preferred level.
29 html xml
30 html xml
31 html xml
Ikegami2
File Ikegami-2Shift-DATA1.ros
Problem Ikegami
Employees 28
Shift types 2
Period 30 days
Cover type Cover is per shift and by skill level.
References [IKE03]
0 html xml
Ikegami3
File Ikegami-3Shift-DATA1.ros
Problem Ikegami
Employees 25
Shift types 3
Period 30 days
Cover type Cover is per shift and by skill level.
References [IKE03]
2 html xml
5 html xml
6 html xml
Other versions Ikegami-3Shift-DATA1.1.ros Some extra personal requests.
3 html xml
9 html xml
13 html xml
Ikegami-3Shift-DATA1.2.ros More requests.
3 html xml
11 html xml
12 html xml
Millar
File Millar-2Shift-DATA1.ros
Problem Millar
Employees 8
Shift types 2
Period 14 days
Cover type Cover is per shift. Over and under coverage is not allowed.
References [IKE03, MIL98]
0 html xml
0 html xml
Other versions Millar-2Shift-DATA1.1.ros A simpler version with less constraints.
0 html xml
BCV3
File BCV-3.46.2.ros
Problem ANROM
Employees 46
Shift types 3
Period 26 days
Cover type Cover is specified per shift.
References [BRU07, BRU09, BUR07b, BUR07c]
894 html xml
895 html xml
896 html xml
BCV4
File BCV-4.13.1.ros
Problem ANROM
Employees 13
Shift types 4
Period 29 days
Cover type Cover is specified per shift.
References [BRU07, BRU09, BUR07b, BUR07c]
10 html xml
Azaiez
File Azaiez.ros
Problem Azaiez
Employees 13
Shift types 2
Period 28 days
Cover type Cover is per shift and by skill level.
References [AZA05]
0 html xml
Valouxis
File Valouxis-1.ros
Problem Valouxis
Employees 16
Schedule length 28 days
Shift types 3
Cover type Cover is specified per shift.
References [VAL00]
20 html xml
40 html xml
60 html xml
120 html xml
HED01
File HED01.ros 
Problem HED01
Employees 20
Schedule length 31 days
Shift types 5
Cover type Cover is per shift and includes skills.
Comments (Uses the Conditional constraint).
References [PUE09]
136 html xml
138 html xml
140 html xml
Other versions HED01b.ros A modified version used in SOL13.
129 html xml
BCDT-Sep
File BCDT-Sep.ros 
Problem BCDT
Employees 20
Schedule length 30 days
Shift types 4
Cover type Cover is per shift.
References [BEL04]
100 html xml
140 html xml
160 html xml
WHPP
File WHPP.ros
Problem WHPP
Employees 30
Schedule length 2 weeks
Shift types 3
Cover type Cover is specified per shift.
References [WEI95]
5 html xml
LLR
File LLR.ros
Problem LLR
Employees 27
Schedule length 7 days
Shift types 3
Cover type Cover is specified per shift.
References [LI03]
301 html xml
366 html xml
Musa
File Musa.ros
Problem Musa
Comments This instance is taken from a fairly early publication on nurse rostering. A UNIVAC 1100 was used to solve it in 1982!
Employees 11
Schedule length 14 days
Shift types 1
Cover type Cover is specified per shift, minimum and preferred levels are used.
References [MUS84]
175 html xml
199 html xml
Ozkarahan
File Ozkarahan.ros
Problem Ozkarahan
Employees 14
Schedule length 7 days
Shift types 2
Cover type Cover is specified per shift.
References [OZK89]
0 html xml
MER
File MER.ros
Problem Montreal
Employees 54
Shift types 12
Period 6 weeks
Cover type Cover is per time period of the day.
7081 html xml
7213 html xml
7248 html xml
CHILD
File CHILD.ros
Problem Montreal
Employees 41
Shift types 5
Period 6 weeks
Cover type Cover is per time period of the day.
149 html xml
154 html xml
155 html xml
ERRVH
File ERRVH.ros
Problem Montreal
Employees 51
Shift types 8
Period 6 weeks
Cover type Cover is per time period of the day.
2001 html xml
2021 html xml
2080 html xml
ERMGH
File ERMGH.ros
Problem Montreal
Employees 41
Shift types 4
Period 6 weeks
Cover type Cover is per time period of the day.
779 html xml
(bold denotes proven optimal solution)

References

  • Asta, S., Özcan, E., and Curtois, T. A tensor based hyper-heuristic for nurse rostering. Knowledge-based systems, 2016. 98: p. 185-199.
  • Burke E.K. and T. Curtois. New Approaches to Nurse Rostering Benchmark Instances. European Journal of Operational Research, 2014. 237(1): p. 71-81. pdf.
  • Solos, Ioannis P., Ioannis X. Tassopoulos and Grigorios N. Beligiannis. A Generic Two-Phase Stochastic Variable Neighborhood Approach for Effectively Solving the Nurse Rostering Problem. Algorithms, 2013. 6: p. 278-308.

See also

  • - A spreadsheet of published results and papers under review.
  • - Lower bounds.
  • - A formal description and model for the BCV instances (a zip file containing all the BCV instances is available here).
  • - The instances in the format used by HyFlex in the Cross-domain Heuristic Search Challenge are available here.