Presentation on theme: "What is MORe? A consulting group based at the Department of Mathematics and Statistics University of Melbourne providing Operations Research (OR) Consulting."— Presentation transcript:
What is MORe? A consulting group based at the Department of Mathematics and Statistics University of Melbourne providing Operations Research (OR) Consulting to business, industry, government & community organisations.
Strategic decision making –resource requirements and allocations e.g. vehicles, machines, operators –location/relocation decisions e.g. building new facilities, relocating/closing existing facilities –system design e.g. warehouse Operational decision making –scheduling –timetabling –sequencing –routing –rostering –production planning –inventory control What are Operations Research Problems?
MORe People Olivia Smith Dr Heng-Soon Gan Prof. Peter Taylor Prof. Bob Johnston Prof Natashia Boland University of Newcastle Prof Mark Wallace Monash University
Mathematical modelling and analysis: –R–Resource planning, including inventory management –S–Strategic and operational planning –S–System design and optimisation –S–Scheduling and timetabling –Y–Yield management Optimisation and simulation systems –L–Linear programming and mixed integer programming –D–Discrete event and macro simulation –C–Constraint programming –Q–Queuing theory and risk analysis Our Expertise
Mathematical Solution Method (Algorithm) Real Practical Problem Mathematical (Optimization) Problem x2x2 Computer Algorithm Human Decision-Maker Decision Support System Mathematics in Operation
Warehouse Design Maximizing Operational Efficiency in collaboration with Agilistics Incoming Products Customer Order History Operations Research Technology Customers Storage Racks Where to store products? …affects order picking Storage location Orders Pick order Savings of 27-43% on order picking costs Warehouse Design
Defence Objectives What types of forces should be maintained? What force strength is required? Force Optimisation Defence Science and Technology Organisation (DSTO), Department of Defence, Australian Government
Performance measure based on tail probability: –no more than 10% of customers wait more than 5 minutes for a bus, in the car park and at the Airport Terminals. Customer Waiting Time Probability distribution function 5 minute Bus Service Objective
A consulting project with Hi Fert Pty Ltd Transport and Logistics
Hi Fert Australian importer and distributor of agricultural materials.Australian importer and distributor of agricultural materials. Operates 9 coastal distribution facilities across the east coast of Australia.Operates 9 coastal distribution facilities across the east coast of Australia. Approx 30% market share (#2 player).Approx 30% market share (#2 player). BRISBANE NEWCASTLE ADELAIDE KADINA PORT LINCOLN PORTLAND GEELONG TOWNSVILLE MACKAY
Ship Route & Trucks Red Sea Suppliers Israel, Egypt BRISBANE NEWCASTLE ADELAIDE KADINA PORTLAND GEELONG TOWNSVILLE MACKAY 2500 3000 5500 2500 1500 16500 1400 3400 10000 7000 Ship Holds Material A:29400 T Material B:6900 T 200 80
The companys old process involved iteratively matching a Shipping Schedule to Raw Material Requirements: –manual and time consuming –predominantly volume-based constraints considered –no robust assurance that the plan was ultimately optimal –poor knowledge management –planning 3 months ahead Existing ERP (SAP) could not cope with the complex maths of the problem. Sales Forecast Raw Material Requirements Shipping Schedule Sourcing Plan OPTIMISATION TOOL Sales Forecast Raw Material Requirements Shipping Schedule Sourcing Plan The desire was to create an Optimisation Tool with all constraints built in to develop the Sourcing Plan with the lowest overall cost. – –12-15 months planning horizon – –centralised information management – –guided feasibility checking – –automated generation of reports The Optimisation Tool
How can orders be combined for practical shipment? Which orders can come forward, which delayed? Should the order in a particular period be from a supplier in a different region in order to facilitate combination shipping? Even if the price is higher? How should the order quantities change to facilitate combination in a shipment? Key Questions
Decisions are interlinked. Need wholistic solution techniques that make all decisions simultaneously, and account for interactions. Linear (integer) programming is the key.Solution
Carbon Liability Optimiser In a carbon constrained environment firms must account for their emissions liability when making strategic planning decisions. Carbon Liability Optimiser (CarLO) formulates a profit maximising carbon strategy while accounting for a firms emissions liability and regulatory constraints.
Formulates an investment strategy that incorporates both carbon and non-carbon related investments across all areas of business. Accounts for the complex interactions between investment activities, the emissions liability, and regulatory requirements. Explicitly accounts for the requirements of the Carbon Pollution Reduction Scheme legislation. Allows quick and simple what-if analysis CarLO…
MORe has been advising a client on stochastic programming approaches for solving a multi-period economic model. Creative decomposition techniques and solution algorithms have been developed to handle the excessive size of the problem. Stochastic Programming
Why Account for Stochasticity? Deterministic programs provide an optimal solution given fixed parameter values. This solution will not be optimal when the parameter values change. Stochastic programming seeks a solution that is optimal over a range of uncertain scenarios. This provides more robust and realistic outcomes. This is particularly valuable in strategic planning.
Expand capacity Take no action High demand High demand Low demand Low demand Decision 1 Consider a capacity expansion decision. The return depends upon uncertain demand. A stochastic program will find a complete strategy of optimal decisions given all demand scenarios. Uncertain Outcome Highest return Low return Moderate return Expand capacity Take no action High demand High demand Low demand Decision 2 High return Low demand Uncertain Outcome An Example…
Facility Location Decisions Operations Research: A What-if Application
36km W-4 Facility Location A A F F D D C C W-1 W-2 W-3 W-5 W-6 Customer Warehouse (W) Assume: Transportation cost: $20/km/unit Transportation cost: $20/km/unit Warehouses have the same O/H cost Warehouses have the same O/H cost Warehouse has very large capacity Warehouse has very large capacity Problem modelled as an integer linear program, and solved using Xpress MP. 10 000 units 180 000 10 000 180 000 220 000 10 000 B B E E 36km
Scenario 1: Warehouse O/H cost is very small as compared to transportation cost –Warehouse O/H: $6 000 000 –Transportation cost: $20/km/unit –proximity dominates –operate the warehouse closest to each customer W-4 A A F F D D C C W-1 W-2 W-3 W-5 W-6 10 000 units 180 000 10 000 180 000 220 000 10 000 B B E E Facility Location
Scenario 2: Warehouse O/H cost is very large as compared to transportation cost –Warehouse O/H: $1 800 000 000 –Transportation cost: $20/km/unit –too expensive to operate a warehouse –hence, the most centralised warehouse selected (based on demand & distance) W-4 A A F F D D C C W-1 W-2 W-3 W-5 W-6 10 000 units 180 000 10 000 180 000 220 000 10 000 B B E E Facility Location
Scenario 3: Both warehouse O/H and transportation costs are competing –Warehouse O/H: $60 000 000 –Transportation cost: $20/km/unit –solution is not obvious; too many possibilities W-4 A A F F D D C C W-1 W-2 W-3 W-5 W-6 10 000 units 180 000 10 000 180 000 220 000 10 000 B B E E Facility Location
Scenario 4: Both warehouse O/H and transportation costs are competing AND warehouse capacity limited –Warehouse O/H: $60 000 000 –Transportation cost: $20/km/unit –Warehouse capacity: 150 000 units W-4 A A F F D D C C W-1 W-2 W-3 W-5 W-6 10 000 units 180 000 10 000 180 000 220 000 10 000 B B E E 70 000 10 000 30 000 110 000 150 000 70 000 10 000 Facility Location