Harold Larnder Prize

Omond Solandt Award

Award of Merit

Service Awards

Practice Prize Competition

Practice Prize Competition

Description

Each year the Canadian Operational Research Society conducts a competition on the practice of operational research to recognize the challenging application of the OR approach to the solution of applied problems. The main criteria considered in evaluating submissions are project impact on the client organization, contribution to the practice of O.R., quality of analysis, degree of challenge and quality of written and oral presentation.

There are two basic purposes behind the Competition, the first being the obvious one of recognizing outstanding O.R. practice. The other is to focus attention on O.R. and its applications by practitioners from Canada by attracting quality papers to the CORS National Conference.

(Top)

Andrea Cameron, Corinne MacDonald Eldon Gunn, Department of Industrial Engineering, Dalhousie University

The Scotsburn Dairy Group is now the largest Atlantic Canadian dairy producer and distributor, producing a variety of products including milk, butter, cottage cheese, sour cream, and frozen dairy desserts. At their Truro Ice Cream facility, the company produces over 300 different products on three production lines, for a total of over 30 million litres of frozen goods products per year.

There are several aspects to ice cream production that make scheduling and inventory a real challenge. First, ice cream demand is highly seasonal; demand during the summer months well exceeds the plant’s capacity to produce. Once produced, ice cream products can be stored for several months, but the cost of such freezer storage is obviously high. The production of ice cream begins with the production of large batches of base mix, which must first be stored for several hours. Then, the mix is piped to the production line where flavours and/or additional ingredients such as nuts are added, and then after passing through a freezer, the ice cream is packaged into containers. In all, the plant produces 30 different base mixes, each of which may be used for more than one final flavour. Another complexity is allergens such as nuts. If a production line runs a product with an allergen, then for the rest of the production day only products with the same allergen can be run on that production line. The lines are thoroughly cleaned during the nightly shutdown.

Each fall, the marketing and production teams would generate a forecast for the upcoming year. Then, in order to deal with the forecasted increase in demand for the summer, the plant would expand its workforce in the spring, adding a second shift and/or working on Saturdays. Anticipatory inventory would be built up so forecasted summer demand could be met. Often this would be started too early and inventory would last well into the fall.

A few days before the beginning of each production week, the production planner would evaluate inventory and forecasted and/or confirmed orders and recommend the SKUs and quantities to be produced. Each SKU had a typical batch size that would be run whenever the product was produced; the planner based these decisions mostly on past history and experience. Once the SKUs and quantities were decided, the final step was to determine the production sequence; again, the planner relied heavily on experience to determine the “best” sequence, keeping the allergen constraint in mind while trying to minimize lost time due to setups (which are highly sequence dependent).

It was apparent that we could apply operations research to all three levels of this problem. Working with the firm’s production management team, models were developed to recommend close to optimal decisions at all three levels of the production planning hierarchy. For the annual level, the goal was to determine staffing levels (and thus production capacity); for the weekly level, the products to be produced in given week and the corresponding quantities; and finally, for the daily level, the best sequence of products on all three lines for each day of the upcoming production week.

The annual aggregate model produces an optimal labour schedule for one year based on minimising holding and labour costs. The model is constrained by the production speed of the lines, the capacity of the freezer and warehouses, and the need to meet demand in every period. Also, rather than allow the model to schedule regular or overtime production at liberty, each month’s schedule must be one of twelve schedules preselected by the company. Constraining the model to predefined schedules has a small negative impact on the computation time but ensures that the results are more practical. The products’ inventory levels and demand forecasts are aggregated based on similar production and storage characteristics. Aggregating the data results in a much faster solution time but does not affect the outcome of the optimal labour scheme at such a high level. The problem is modelled as a linear programming problem and solves to within 0.01% of optimal in less than two minutes. The result is an optimal labour scheme by month for the next twelve months. The model also calculates the associates production and inventory levels needed to meet demand in each period.

The second model in the hierarchy produces a weekly production schedule that minimizes holding costs and setup costs while also meeting demand for the next 13 weeks and adhering to the labour schedule selected by the first model. Products are grouped into families based on shared setup costs to decrease the solution time and also so families of products are scheduled in the same period to save on setups. For each product family, the choices for production quantities are limited to those quantites that will exactly satisfy an integer number of upcoming demand periods. The costs of these production options are calculated and the resulting production and cost matrices are passed to the model. The mixed integer program then minimizes production costs while ensuring all periods are covered and the labour schedule is followed. The resulting family production schedule can be transformed back into individual product runs and the final output is a list of all products scheduled for production next week and their optimal run lengths.

Lastly, the third detailed production model minimizes product loss and lost production time through optimal production sequencing. The available labour and the products to be produced are generated from the earlier models. Constraints include the allergen constraint discussed earlier, production speeds, and several other production constraints. The result is a suggested schedule for production on all three lines in the plant for those products to be produced in the upcoming week.

The first model is fully implemented and has been in use at the firm for some time, and it will play a big part in the firm’s planning for the upcoming year. The second model has been successfully implemented at the facility. Experience to date has been positive, and the model continues to be adapted by adding more complexities of the production environment. The final daily model has been tested on actual company data with positive results. Implementation of the final model is anticipated in the new year.

John Blake and Vinh Quan with Andrea Cameron,
Corinne MacDonald, and Eldon Gunn, winners of the 2009 CORS Practice Prize

Previous Recipients

(Top)

2009

First Prize

Andrea Cameron, Corinne MacDonald Eldon Gunn, "Improving Production and Inventory Planning for Scotsburn Ice Cream"

Second Prize

Sonya Vanderby, Michael W. Carter, Christopher Feindel, Timothy Latham, "Modelling the Future of the Canadian Cardiac Surgery Workforce Using Systems Dynamics"

2008

First Prize

Vincent S. Chow, Derek Atkins, Wenhai Huang, Martin L. Puterman, Neda Salehirad, "Reducing Surgical Ward Congestion at the Vancouver Island Health Authority through Improved Surgical Scheduling"

2007

First Prize

Mustapha Ouhimmou, Sophie D'Amours, Daoud Ait-kadi, Robert Beauregard, Satyaveer Singh Chauhan,"Optimization helps Shermag Gain Competitive Edge"

2006

First Prize

Saeed Zolfaghari, Vinh Quan, Ahmed El-Bouri, Banafsheh Namiranian, Maryam Khashayardoust, "Heuristics for large scale labour scheduling problems in retail sector"

Second Prize

Mehmet Aydinel, Ximena Cerda, Eric Cope, Mats Gerschman, Taraneh Sowlati, "Optimization of Production Allocation and Transshipment for a Leading Canadian Forest Products Company"

Third Prize (2)

Bao Nguyen, David Hopkin, Dr. Handson Yip, Dr. John Fawcett, "Concepts of Operations for the Side Scan Sonar Autonomous Underwater Vehicles Developed at DRDC Atlantic"

Eric Cope, Ian Fichtenbaum, Neda Salehirad, "Optimizing Cargo Heating Practices at Teekay Shipping Corporation"

2005

First Prize

Li Chen, Mats Gerschman, Fredrik Odegaard, David K. Puterman, Martin L. Puterman, Ryan Quee, "Improving the Efficiency of Porter Operations in Two Vancouver Hospitals"

2004

First Prize

Bernard Gendron, “Scheduling Employees in Quebec’s Liquor Stores with Integer Programming”

2003

First Prize

Derek Atkins, Mehmet A. Begen, Bailey Kluczny, Anita Parkinson, “Reducing Pre-Board Screening Queues at Vancouver International Airport”

Second Prize

Brunilde Sansò, Frédéric Bodevin, “Assessing User QoS in a Deployed GPRS Network”

2002

First Prize

Ernest Urbanovich, Ella E.Young, Martin L. Puterman, Sidney O. Fattedad, "Early Detection of High Risk Claims at the Worker's Compensation Board of British Columbia"

Second Prize

Marko Blais, Sophie D.Lapierre, Gilbert Laporte, "Solving a Home Care Districting Problem in an Urban Setting"

Third Prize

A. Ingolfsson, E. Erkut, S. Budge, "Simulation of a Single Start Station for Edmonton EMS"

2001

First Prize

Jean-Marie Bourjolly, Leslie Dejoie, Ke Ding, Oumar Dioume, Michel Lominy, "DOCAF (TM): A Commercial Software Package for Frequency Allocation in Cellular Phone Networks"

Second Prize

Erhan Erkut, Steve Kabanuk, Russ Fenske, Jerry Davis, "Improving the Emergency Service Delivery in St-Albert"

2000

Honourable Mention

Bao Nguyen, Dr Kevin Ng, "Modeling Canadian Search and Rescue Operations"

1999

Claude Berger, Fabien Chauny, André Langevin, Richard Loulou, Diane Riopel, Gilles Savard, Jean-Philippe Waaub, Agnès Wizere, "EUGÈNE : Un système d'aide à la décision pour la planification à long terme de la gestion intégrée des déchets domestiques"

1998

Erhan Erkut, Anthony Myroon, Kevin Strangway, "TransAlta redesigns its service delivery network"

1997

Paul Comeau, "DSS TANKER - A fuel management model for Air Transport Group"

1996

Michel Gamache, François Soumis, Daniel Villeneuve, Jacques Desrosiers, Éric Gélinas, "The preferential bidding problem at Air Canada"

1995

Maurice Elliott, "Optimizing gas production allocation"

1994

John Blake, Mike Carter, "An analysis of emergency room wait time issues via computer simulation"

1993

Michel Gamache, François Soumis, Gerald Marquis, Jacques Desrosiers, "A column generation approach for large scale aircrew rostering problems"

1992

Michael Carter, Gilbert Laporte, "A general examination scheduling system"

1991

Michael Scriabin, Jason Farlette, Dilip Kotak, Michael A. Matthews, "Air capacity planning symbiotic system (AIRSYM): Combining artificial and human intelligence"

1990

Jacques Ferland, Charles Fleurant, "Computer aided scheduling for a sport league"

1989

Eldon A. Gunn, Glenn Allen, John C. Campbell, Barclay Cunningham, Peter Rutherford, "One Year of OR: Models for Operational and Production planning in the Coal Industry"

1988

First Prize

Michael Carter, “A Lagrangian Relaxation Approach to the Classroom Assignment Problem”

Second Prize

Jacques A. Ferland and Gilles Guénette, “Decision Support System For the School Districting Problem”

1987

Not awarded

1986

Gilbert Laporte, "A System for Assigning Students to Course Sections at the École Polytechnique de Montréal"

1985

Not awarded

1984

A tie

Michel Goulet, "Système de Prévision des Déboursés"

Cary Swoveland, "Risk Analysis of Regulatory Options for the Transport of Dangerous Commodities by Rail"

1983

Iain R. Grunwell, "Simulation for Training Railway Yard General Yardmasters"

1982

Not awarded

Practice Prize Application Process

(Top)

Eligibility

Entries will be expected to report on a completed, practical application, and must describe results that had significant, verifiable and preferably quantifiable impact on the performance of the client organization.

Instructions

• Submit an abstract both by e-mail and in hardcopy, not exceeding 500 words of a paper on an actual success story of operational research by March 19, 2010 to:
  
  John Blake
  Department of Industrial Engineering
  Dalhousie University
  Halifax, NS B3J 2X4
  Phone: (902) 494-6068
  Fax: (902) 420-5878
  E-mail: john.blake@dal.ca
  
  The work on the project may have taken place over a period of several years, but at least some of the work must have taken place over the last two years. Previous publication of the work does not disqualify it; however, you may not report on a project which has been previously submitted to the CORS Competition on the Practice of O.R.
  
  
• Be a resident of Canada.
• Include a letter by an executive of the client organization that sponsored the application, attesting that this application truly had an impact and that the organization would not object to having a paper presented.
• Include the phone number of the author(s) and the name(s), title(s) and phone number(s) of at least one executive of the client organization where the O.R. application was put into effect.
• Finalists will be selected by April 9, 2010. They must submit a more detailed written report by April 30, 2010, which must not exceed 25 pages in the body and make an oral presentation of the paper at the CORS Annual Conference in Edmonton, AB, May 25-29, 2010. Finalists are required to prepare a one page project summary, suitable for publication in the CORS Bulletin and on the CORS website, detailing their project and its impact on the client organization.

Selection Criteria

• The project should exemplify the challenging application of the operational research approach to the solution of significant applied problems. While statistical and presentational difficulties are characteristics of almost all applications, they will not be considered a substitute for operational research.
• The quality of the analysis, of the modeling and of the successful implementation of the results at the client organization.
• The significance of the impact of the results and recommendations on the performance of the client organization.
• The presentation, both written and oral. A crucial part of the presentation is a "case history" of the project, which describes the project's development from start to finish, and focuses upon the challenges faced by the analysts.

Awards

A total of $1800 may be awarded by the committee: however, the committee reserves the right not to award any prize.