32
Scenario Trees and Metaheuristics for Stochastic Inventory Routing Problems DOMinant Workshop, Molde, Norway, September 21.-22., 2009 Lars Magnus Hvattum Norwegian University of Science and Technology, Trondheim, Norway Arne Løkketangen Molde University College, Molde, Norway Gilbert Laporte HEC, Montréal, Canada

Scenario Trees and Metaheuristics for Stochastic Inventory Routing Problems

  • Upload
    nicola

  • View
    26

  • Download
    0

Embed Size (px)

DESCRIPTION

Scenario Trees and Metaheuristics for Stochastic Inventory Routing Problems. Lars Magnus Hvattum Norwegian University of Science and Technology , Trondheim, Norway Arne Løkketangen Molde University College, Molde, Norway Gilbert Laporte HEC, Montréal, Canada. - PowerPoint PPT Presentation

Citation preview

Page 1: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

Scenario Trees and Metaheuristics forStochastic Inventory Routing Problems

DOMinant Workshop, Molde, Norway, September 21.-22., 2009

Lars Magnus HvattumNorwegian University of Science and Technology, Trondheim, Norway

Arne LøkketangenMolde University College, Molde, Norway

Gilbert LaporteHEC, Montréal, Canada

Page 2: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

2Outline of the presentation

Page 3: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

3Inventory Routing Problems extend the VRP, and look at a larger part of a supply chain

Supplier Customer

VR

PIR

P

Page 4: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

4The dynamic aspect of inventory routing can be handled in various ways

1) Single Period models 2) Multi-Period models

3) Infinite horizon models

(...) (...)

Kleywegt, Nori, and Savelsbergh (2002, 2004)

Adelman (2004)

Page 5: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

5The problem is modelled as a Markov Decision Process, where each epoch corresponds to a day

Page 6: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

6The problem is modelled as a Markov Decision Process, where each epoch corresponds to a day

Page 7: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

7The infinite horizon discounted reward Markov Decision Problem can only be solved for tiny instances

A large number of states: all possible inventory levels

A large number of actions: all possible delivery patterns

A large number of transitions: all possible demand realizations

(...)

(...)

(...)

Page 8: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

8Previous solution methods for these SIRPs are based on approximating the value functions

Standard algorithms for MDPs can solve some instances: • 4 customers,1 vehicle, and 9 inventory levels• 5 customers, 5 vehicles, 5 inventory levels, but only direct

delivery

Kleywegt et al. propose approximations of the value function for instances with direct delivery or at most 3 deliveries per route

Adelman proposes approximations of the value function for instances with an unlimited number of vehicles

Page 9: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

9To simplify, we only look at finite scenario trees to generate a stochastic policy

The infinite horizon is approximated by a finite tree

The large number of transitions is approximated by sampled realizations

The action space remains unchanged

State transitions are based on current inventory level, delivered quantities and sampled demand realizations

Page 10: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

10The problem of finding optimal actions conditional on the sampled tree is formalized as an integer program

Page 11: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

11We have examined three methods to solve the scenario tree problem (STP)

CPLEX• Solve IP• Solve MIP, where integrality constraints are kept only for root node

GRASP• Construct solutions to the STP in a randomized adaptive fashion,

but with added learning mechanisms

Progressive Hedging Algorithm• Decompose problem over scenarios, solve each scenario using

modified GRASP• Gradually enforce an implementable solution by penalties, including

a quadratic term

Page 12: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

12

u = 0,2,0,0

y = 0,0,0,0

d = 1,2,2,1d = 1,1,2,2

y = 0,0,0,0 y = 0,0,0,0

u = 0,0,0,0u = 0,1,0,0

In GRASP, we start with a solution without any deliveries, then add more deliveries if profitable

(...) (...) (...) (...)

No deliveries scheduled in any

node

Page 13: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

13

u = 0,2,0,0

y = 0,0,2,0

d = 1,2,2,1d = 1,1,2,2

y = 0,0,0,0 y = 0,0,0,0

u = 0,0,0,0u = 0,1,0,0

In GRASP, we start with a solution without any deliveries, then add more deliveries if profitable

(...) (...) (...) (...)

First iteration: add 2 units of

delivery to customer 3 using

vehicle 1 in node 1

No increase in inventory, but

reduction in stock-outs

Page 14: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

14

u = 0,2,0,0

y = 0,0,2,2

d = 1,2,2,1d = 1,1,2,2

y = 0,0,0,0 y = 0,0,0,0

u = 0,0,0,1u = 0,1,0,0

In GRASP, we start with a solution without any deliveries, then add more deliveries if profitable

(...) (...) (...) (...)

Second iteration: add 2 units of

delivery to customer 4 using

vehicle 1 in node 1

Increase in inventory in node 3

for customer 4

Page 15: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

15

u = 0,2,0,0

y = 0,0,2,2

d = 1,2,2,1d = 1,1,2,2

y = 0,0,2,0 y = 0,0,0,0

u = 0,0,0,1u = 0,1,0,0

In GRASP, we start with a solution without any deliveries, then add more deliveries if profitable

(...) (...) (...) (...)

Continue making insertions: y

additional units to customer i using

vehicle k in node v

Stop when no profitable, feasible insertions can be

made

Page 16: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

16Several variations of the GRASP are examined, to find a robust version that can handle different types of instances

• Restricted Candidate List (RCL) based on either value or rank

[former is more robust]• Size of RCL is controlled by a parameter that is adjusted dynamically

[more robust than a fixed value]• Build solution node by node (recursively) or all nodes simultaneously

[former is much faster]• Use learning based on analysing completed solutions to find potential

improvements

[increases robustness of building solution node by node]

Page 17: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

17The progressive hedging algorithm is based on decomposing the problem over scenarios

Page 18: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

18Single scenarios are solved separately, until implementability is enforced through penalties

Each scenario is solved using GRASP

Objective function is modified to penalize deviations from an averaged solution:

Page 19: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

19Several variations of the progressive hedging algorithm are examined, to find a robust version that can handle different types of instances

• A progressive hedging heuristic is employed to generated feasible solutions to the original scenario tree problem

[ensures that good feasible solutions are found even if the method has insufficient time to converge]

• Penalty parameter is updated dynamically

[better balance of progress towards an implementable solution]• Use multiple penalty parameters and weights

[better than using a single parameter]• Use intermediate heuristic solutions for guiding the search

[does not work, quicker convergence but to worse solutions]• Lock variables for which concensus seems to have been reached

[does not work, induces cycling behavior?]

Page 20: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

20The search of the resulting progressive hedging algorithm has a fairly similar behavior across instances

Left: the averaged solution gives a heuristic upper bound, and the progressive hedging heuristic gives actual lower bounds

Right: we can measure the distances in the solution space and the parameter space between iterations, as well as the dynamic penalty parameter

Page 21: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

21There are different ways of controlling the computational effort used by the methods

Increasing the effort gives improved results

Right: profit as a function of the number of GRASP iterations per epoch

Page 22: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

22The size of the scenario trees is crucial both for the computational time and the simulation results

Increasing the size of the scenario trees increases the computational effort as well as the profits observed

Page 23: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

23Several scenario tree problems were studied separately to study the GRASP, the PHA, and the other methods

Name LP(-STE) DS Exact IP4 MIP TDG ANGPHA:TDG

PHA:ANG

STP 01 720.2 - 651.0 687.1 691.0 690.5 690.5 692.7 693.0STP 04 798.2 - 724.0 753.7 756.7 757.1 757.2 754.2 755.3STP 05 157626.0 - - - 73652.9 151958.0 151378.0 146832.0 139283.0STP 07 157352.0 - - 126497.0 148095.0 151960.0 151458.0 151458.0 143851.0STP 11 -266.8 -617.7 - - - -473.3 -473.4 -478.7 -461.9STP 15 -420.7 -869.4 - - - -793.3 -772.9 -763.1 -748.9STP 16 - -984.7 - - - -938.1 -909.4 -933.3 -925.0STP 21 -270.2 -623.9 - - - -479.7 -478.7 -487.5 -470.4STP 26 -113.3 -909.2 - - - -563.4 -514.0 -518.8 -536.7STP 31 67452.7 - - - 62012.1 62678.4 62620.6 63283.6 63336.2STP 41 4570.6 - 4137.8 - 4195.1 4111.6 4181.3 4116.0 4161.1STP 42 - - - - - -160.5 89.5 -900.1 -1278.2STP 43 1563.0 - - - - -1155.3 -1149.9 -1119.1 -1683.9STP 51 33309.4 - - 32537.7 32537.7 32365.2 32365.2 32365.2 32365.2STP 52 22268.1 - - 21740.9 21740.9 21433.7 21433.7 21411.7 21411.7STP 55 1375.0 - 1071.9 1232.6 1213.5 1229.9 1229.9 1231.7 1232.6

Page 24: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

24Simulations are run over many epochs (600), and quick heuristics for the scenario tree problems must be selected

Our methods: no initial time required, but should allocate some time for the daily problem (time consuming when evaluating, but ok in practice)

Other methods: high effort initially (days), but fairly quick for the daily problem (ok when evaluating, practice requires a stable situation)

Page 25: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

25GRASP is quicker than PHA, but PHA is better on some instances (produces solutions with different structure)

Page 26: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

26Some of the observations made during the work have potential for leading to future research

1) Scenario trees are used to represent the stochastic and dynamic aspects: how to incorporate these in the most efficient way?

2) Scenario tree problems must be solved to generate decisions: how is this best done?

Page 27: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

27Scenario trees are not often used in (meta-) heuristics, and several questions remain as to how they should be generated

We generate them using random sampling to cover a specified tree structure

The size of the tree is determined by specifying the branching factors for each level of the tree

Page 28: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

28Can we save computational time by using a clever tree structure?

We use lower branching factors lower in the tree (more important to represent stochasticity that is close in time?)

We should vary the depth/width of the tree based on the instance solved?

We use the same length for every scenario

We had to limit the size of the tree to be able to evaluate the methods with simulations

Page 29: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

29We can improve results for a given tree size by sampling differently?

We also tried a moment matching method, but a requirement of the implementation was that for a single parent node, the number of children must be at least the number of customers (distributions)

Research question: can we determine a suitable objective function to be used by a metaheuristic that constructs scenario trees?

We use random sampling: by using larger trees we get closer to the true distribution

Page 30: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

30We have tested GRASP and PHA for solving the scenario tree problems

Local search based methods frequently perform better and faster than GRASP

Problem: difficult to find local moves in a scenario tree, as the interconnectedness of decisions creates feasibility issues

(how to find a suitable move evaluation? how to guide the search back to feasible space if allowing capacity/inventory violations?)

Page 31: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

31One idea is to hybridize local search and construction heuristics

Solution:

Do local search moves only in this part of the solution (representation)

To evaluate each move, this part of the solution must be completed using a construction heuristic

(Root node) (Rest of tree)

Page 32: Scenario Trees and  Metaheuristics  for Stochastic Inventory Routing Problems

32Concluding remarks

Stochastic and dynamic problems may become increasingly important (with better technology and access to data)

The stochastic inventory routing problem is an interesting playground for testing how one can deal with stochastic and dynamic problems

Using scenario trees to represent stochasticity is relatively untested in combination with metaheuristics

Several directions for future research have been found but not yet pursued