Skip to main content
Log in

Tackling a VRP challenge to redistribute scarce equipment within time windows using metaheuristic algorithms

EURO Journal on Transportation and Logistics

Cite this article


This paper reports on the results of the VeRoLog Solver Challenge 2016–2017: the third solver challenge facilitated by VeRoLog, the EURO Working Group on Vehicle Routing and Logistics Optimization. The authors are the winners of second and third places, combined with members of the challenge organizing committee. The problem central to the challenge was a rich VRP: expensive and, therefore, scarce equipment was to be redistributed over customer locations within time windows. The difficulty was in creating combinations of pickups and deliveries that reduce the amount of equipment needed to execute the schedule, as well as the lengths of the routes and the number of vehicles used. This paper gives a description of the solution methods of the above-mentioned participants. The second place method involves sequences of 22 low level heuristics: each of these heuristics is associated with a transition probability to move to another low level heuristic. A randomly drawn sequence of these heuristics is applied to an initial solution, after which the probabilities are updated depending on whether or not this sequence improved the objective value, hence increasing the chance of selecting the sequences that generate improved solutions. The third place method decomposes the problem into two independent parts: first, it schedules the delivery days for all requests using a genetic algorithm. Each schedule in the genetic algorithm is evaluated by estimating its cost using a deterministic routing algorithm that constructs feasible routes for each day. After spending 80 percent of time in this phase, the last 20 percent of the computation time is spent on Variable Neighborhood Descent to further improve the routes found by the deterministic routing algorithm. This article finishes with an in-depth comparison of the results of the two approaches.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12

Similar content being viewed by others







  6. At the time of writing this article, after the all-time-best challenge has ended.


  • Ahmed L, Mumford C, Kheiri A (2019) Solving urban transit route design problem using selection hyper-heuristics. Eur J Oper Res 274(2):545–559

    Article  Google Scholar 

  • Asta S, Özcan E (2015) A tensor-based selection hyper-heuristic for cross-domain heuristic search. Inf Sci 299:412–432

    Article  Google Scholar 

  • Battarra M, Cordeau JF, Iori M (2014) Pickup-and-delivery problems for goods transportation. In: Toth P, Vigo D (eds) Vehicle Routing, chap 6, pp 161–191. doi:

    Chapter  Google Scholar 

  • Burke EK, Gendreau M, Hyde M, Kendall G, Ochoa G, Özcan E, Qu R (2013) Hyper-heuristics: a survey of the state of the art. J Oper Res Soc 64(12):1695–1724

    Article  Google Scholar 

  • Clarke G, Wright JW (1964) Scheduling of vehicles from a central depot to a number of delivery points. Oper Res 12(4):568–581

    Article  Google Scholar 

  • Coelho LC, Cordeau JF, Laporte G (2014) Thirty years of inventory routing. Transp Sci 48(1):1–19

    Article  Google Scholar 

  • Cordeau JF, Laporte G (2007) The dial-a-ride problem: models and algorithms. Ann Oper Res 153(1):29–46

    Article  Google Scholar 

  • Croes GA (1958) A method for solving traveling-salesman problems. Oper Res 6(6):791–812

    Article  Google Scholar 

  • Dethloff J (2001) Vehicle routing and reverse logistics: the vehicle routing problem with simultaneous delivery and pick-up. OR-Spektrum 23(1):79–96

    Article  Google Scholar 

  • Dror M, Fortin D, Roucairol C (1998) Redistribution of self-service electric cars: a case of pickup and delivery. Tech. Rep. RR-3543, INRIA

  • Dueck G (1993) New optimization heuristics: the great deluge algorithm and the record-to-record travel. J Comput Phys 104(1):86–92

    Article  Google Scholar 

  • Dullaert W, Gromicho J, van Hoorn J, Post G, Vigo D (2017) The VeRoLog solver challenge 2016–2017. J Veh Rout Algorithms 1:1–3.

    Article  Google Scholar 

  • Fagerland MW, Sandvik L (2009) The Wilcoxon–Mann–Whitney test under scrutiny. Stat Med 28(10):1487–1497

    Article  Google Scholar 

  • Fishman E (2016) Bikeshare: a review of recent literature. Transp Rev 36(1):92–113

    Article  Google Scholar 

  • Gromicho JA, Haneyah S, Kok L (2015) Solving a real-life vrp with inter-route and intra-route challenges.

  • Hansen P, Mladenović N (1999) An introduction to variable neighborhood search. In: Meta-heuristics, Springer, pp 433–458

  • Hart E, Ross P, Corne D (2005) Evolutionary scheduling: a review. Genetic Progr Evol Mach 6(2):191–220

    Article  Google Scholar 

  • Jian N, Freund D, Wiberg HM, Henderson SG (2016) Simulation optimization for a large-scale bike-sharing system. In: Proceedings of the 2016 Winter Simulation Conference, IEEE Press, Piscataway, NJ, USA, WSC ’16, pp 602–613

  • Johnson DS (1973) Near-optimal bin packing algorithms. PhD thesis, Massachusetts Institute of Technology

  • Kheiri A, Keedwell E (2015) A sequence-based selection hyper-heuristic utilising a hidden Markov model. In: Proceedings of the 2015 on genetic and evolutionary computation conference, GECCO ’15, pp 417–424

  • Kheiri A, Keedwell E (2017) A hidden markov model approach to the problem of heuristic selection in hyper-heuristics with a case study in high school timetabling problems. Evolut Comput 25(3):473–501

    Article  Google Scholar 

  • Kheiri A, Keedwell E, Gibson MJ, Savic D (2015) Sequence analysis-based hyper-heuristics for water distribution network optimisation. Procedia Engineering 119:1269–1277, computing and Control for the Water Industry (CCWI2015) Sharing the best practice in water management

    Article  Google Scholar 

  • Kruskal WH (1957) Historical notes on the Wilcoxon unpaired two-sample test. J Am Stat Assoc 52(279):356–360

    Article  Google Scholar 

  • Lin S, Kernighan BW (1973) An effective heuristic algorithm for the traveling-salesman problem. Oper Res 21(2):498–516

    Article  Google Scholar 

  • Martello S, Toth P (1990) Lower bounds and reduction procedures for the bin packing problem. Discrete Appl Math 28(1):59–70

    Article  Google Scholar 

  • Masson R, Lehuede F, Peton O (2014) The dial-a-ride problem with transfers. Comput Oper Res 41:12–23

    Article  Google Scholar 

  • Min H (1989) The multiple vehicle routing problem with simultaneous delivery and pick-up points. Transp Res Part A: General 23(5):377–386.

    Article  Google Scholar 

  • Montané FAT, Galvão RD (2006) A tabu search algorithm for the vehicle routing problem with simultaneous pick-up and delivery service. Comput OR 33:595–619

    Article  Google Scholar 

  • Parragh S, Doerner K, Hartl R (2008) A survey on pickup and delivery problems: Part II: Transportation between pickup and delivery locations. Journal für Betriebswirtschaft 58:81–117.

    Article  Google Scholar 

  • Pfrommer J, Warrington J, Schildbach G, Morari M (2014) Dynamic vehicle redistribution and online price incentives in shared mobility systems. IEEE Trans Intell Transp Syst 15(4):1567–1578

    Article  Google Scholar 

  • Rainer-Harbach M, Papazek P, Hu B, Raidl GR (2013) Balancing bicycle sharing systems: a variable neighborhood search approach. In: Middendorf M, Blum C (eds) Evolutionary computation in combinatorial optimization. Springer, Berlin, pp 121–132

    Chapter  Google Scholar 

  • Rand GK (2009) The life and times of the savings method for vehicle routing problems. ORiON 25(2):125–145

    Article  Google Scholar 

  • Raviv T, Kolka O (2013) Optimal inventory management of a bike-sharing station. IIE Trans 45(10):1077–1093

    Article  Google Scholar 

  • Schilde M, Doerner K, Hartl R (2011) Metaheuristics for the dynamic stochastic dial-a-ride problem with expected return transports. Comput Oper Res 38(12):1719–1730

    Article  Google Scholar 

  • Schuijbroek J, Hampshire R, van Hoeve WJ (2017) Inventory rebalancing and vehicle routing in bike sharing systems. Eur J Oper Res 257(3):992–1004

    Article  Google Scholar 

  • Sörensen K, Glover FW (2013) Metaheuristics. In: Gass SI, Fu MC (eds) Encyclopedia of operations research and management science. Springer, Berlin, pp 960–970

    Chapter  Google Scholar 

  • Wall MB (1996) A genetic algorithm for resource-constrained scheduling. PhD thesis, Massachusetts Institute of Technology

  • Wilson D, Rodrigues S, Segura C, Loshchilov I, Hutter F, Buenfil GL, Kheiri A, Keedwell E, Ocampo-Pineda M, Özcan E, Pena SIV, Goldman B, Rionda SB, Hernandez-Aguirre A, Veeramachaneni K, Cussat-Blanc S (2018) Evolutionary computation for wind farm layout optimization. Renew Energy 126:681–691

    Article  Google Scholar 

Download references


We express our gratitude to the VeRoLog board as well as the organizing committee for the VeRoLog Conference that was held in Amsterdam, Netherlands, July 10–12, 2017. The last three authors wish to thank Gerhard Post and Daniël Mocking for co-organizing the VeRoLog Solver Challenge 2017. Alina G. Dragmir and David Mueller (team ADDM) have achieved their results for the all-time-best challenge using the Vienna Scientific Cluster. Additionally, Alina G. Dragomir would like to gratefully acknowledge the financial support by FWF the Austrian Science Fund (Project number P 27858).

Author information

Authors and Affiliations


Corresponding author

Correspondence to Ahmed Kheiri.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix A: Challenge rules

Appendix A: Challenge rules

We summarize the challenge rules, which were originally published in (Dullaert et al. 2017). The challenge consisted of three parts, and the first two ran partially parallel in time.

1.1 Appendix A1: All-time-best challenge

The organizers disclosed 25 instances in December 2016: the “all-time-best instances”. Participants were invited to submit a solution to an instance if it was better than the best solution submitted so far for this instance. Progress, i.e., the cost of the best solution over time, was shown to the participants. This information is still visible on the website, and it shows that different instances are won by different participants.

The all-time-best challenge ran till July 2, 2017 and the participants were rewarded in two ways: for every week during the all-time-best period that their solution was the best, and additionally for having the best solution at the end of the challenge. In this part of the challenge, any means, resources and time, were allowed.

1.2 Appendix A2: Restricted resources challenge

This challenge had a more “traditional” form: the resources were restricted, especially the computing time. The time \(T_{\text {limit}}\) (seconds) that each algorithm was allowed to run on the organizers’ single core machine is limited by the formula \(T_{\text {limit}}=10+2|R|\). Here |R| is the number of (delivery) requests in the instance. The organizers provided a calibration tool, so that each participant could estimate the equivalent time on his or her local machine. In addition, it was not allowed to use any software that is not freely available for commercial use. In particular, this means that for example the use of commercial MILP solvers was forbidden. Each algorithm had to run on a new set of 25 instances (available since April 1 2017). Furthermore, each solver had to run on each instance, using nine different random seeds, to reduce the variance coming from randomized algorithms. Non-randomized algorithms could also profit from the random seeds: they were known to be between \(10^8\) and \(10^9\) with a different starting digit for each seed, and hence it was possible to detect this and run 9 different deterministic algorithms. The corresponding results and solver binaries were submitted on May 8, 2017.

The evaluation of algorithms in the restricted resources challenge was done as follows. A rank score was calculated per instance for each solver. First, per instance, the two best solutions and the two worst solutions found by the solver were removed. The remaining five solutions were used to compute the score of the solver. If these five solutions were all feasible, their average counted as the score of the solver. Alternatively, if there were infeasible solutions among the middle 5 solutions, that solver was first ranked with respect to the number of feasible solutions, and secondary by the average cost of the feasible ones. Finalists were announced on June 1.

1.3 Appendix A3: The finals

The finalists’ solvers were run by the organizers on a set of 50 not previously disclosed (the so-called hidden) instances. Again, per solver per instance but equal for each finalist, nine runs with different random seeds were done, again with nine different starting digits. A solver ranking per instance was made with the same rules as above, as well as a ranking of the solvers based on these scores. The winner of the challenge was the participant whose solver had the lowest mean of the ranks.

Rights and permissions

Reprints and Permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Kheiri, A., Dragomir, A.G., Mueller, D. et al. Tackling a VRP challenge to redistribute scarce equipment within time windows using metaheuristic algorithms. EURO J Transp Logist 8, 561–595 (2019).

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: