Fdp session rtu session 1
- 1. Faculty Development Program on Operations Management 23-27 November, 2020 Supply Chain Analytics: An Overview Prof. Surya Prakash Singh PhD (IIT Kanpur), PDF (NUS Singapore, MIT USA) Dhananjaya Chair Professor Department of Management Studies Indian Institute of Technology Delhi E-Mail: surya.singh@gmail.com
- 2. Origin of SCM? Entered in public domain by Keith Oliver, a consultant at Booz Allen Hamilton while in an interview for Financial Times in 1982. The term was slow to take hold. But, gained popularity in mid-1990s & a lot of research papers, magazine articles and books came out on the subject. In the late 1990s, it rose to prominence as a management buzzward & operations managers started to use it in their titles with increasing regularity
- 5. According to Council of Supply Chain Management Professionals
- 6. Types ofAnalytics 3 This we will be focussing
- 7. Data and Decision Sciences = Data + Analytics = DATAANALYTICS 4
- 8. Data and Decision Sciences = Big Data + Analytics = BIG DATAANALYTICS 5
- 9. Source: Gilvan C. Souza, Supply chain analytics, Business Horizons (2014) 57, 595—605 6
- 10. What is Decision Science? It is a collection of techniques used for decision-making at the individual and organisational levels such as: • cost-benefit analysis, Harvard University, Centre for Decision Sciences • constrained optimization, • simulation modeling, • behavioral decision theory, • statistical inference, • management control, 7
- 11. Decision Science is also known as a field of study that uses computers, statistics, and mathematics to solve business problems. 8
- 12. How is decision science different from other approaches? Most of the approaches focus on producing new feasible choices that are acceptable, but, decision science is only concerned with making an optimal choice based on available information. 9
- 13. Where is decision science used? Decision science has been used in • business and management, • law and education, • environmental regulation, • military science, • public health and • public policy etc. 10
- 14. Decision science models: Types of decision science models: – Mental (arranging furniture) – Visual (blueprints, road maps) – Mathematical (this we’ll be focussing) 14
- 15. Classification of Decision Science Models: o Quantitative Model o Qualitative Model o Simulation Model AI o Heuristics/ Meta-heuristics Model o Hyper-heuristic Model 15
- 16. o Quantitative Decision Science Models Operations Research Models L.P./N.L.P. (Single Obj.) & Goal Prog Models (Multi Obj.) Binary/ Integer/Mixed Integer models Static/ Dynamic models Stochastic/ Probabilistic models Statistical (Forecasting) Models Financial Decision Models Marketing/HR Decision Models Supply Chain Analytics (This we will be focussing) Use of LINGO for decision science 16
- 17. o Qualitative Decision Making Models Delphi method ISM AHP ANP TOPSIS DEMATEL IRP PROMETHEE ELECTRE EATWOS HYBRID models Etc. 17
- 18. o Simulation Based Decision Models Monte Carlo model Continuous Simulation model Discrete (event) based model Combined Discrete/Continuous model Choice of simulation model is a function of the characteristics of the system and the objectives of the problem. 18
- 19. o High End Decision Science Models Genetic Algorithm Artificial Neural Network Fuzzy Logic Simulated Annealing Tabu Search Ant Colony Cuckoo Algorithm Fire flies algorithm Evolutionary method Integrated approaches . 19
- 20. o Advanced Decision Making Models: Hyper-heuristic Methods “Heuristics that choose heuristics” High level heuristics: It was first coined in 1997 by Jörg Denzinger, Matthias Fuchs and Marc Fuchs. They used it to describe a protocol that chooses and combines several AI methods. Meta-heuristics Choice Function Ant Algorithm Genetic Algorithm … Low level heuristics: different moving strategies, constructive heuristics … 20
- 21. Few examples of decision science application….. Motorola – Procurement of goods and services account for 50% of its costs – Developed an Internet-based auction system for negotiations with suppliers – The system optimized multi-product, multivendor contract awards – Benefits: $600 million in savings 11
- 22. Waste Management – Leading waste collection company in North America – 26,000 vehicles service 20 million residential & 2 million commercial customers – Developed vehicle routing optimization system – Benefits: Eliminated 1,000 routes Annual savings of $44 million 12
- 23. Hong Kong International Terminals – Busiest container terminal in the world – 122 yard cranes serve 125 ships per week – Thousands of trucks move containers in & out of storage yard – Used DSS to optimize operational decisions involving trucks, cranes & storage locations – Benefits: 35% reduction in container handling costs 50% increase in throughput 30% improvement in vessel turnaround time 13
- 24. (Introduction to Prescriptive Analytics: Linear Programming (Operations Research) Techniques)
- 25. Introduction to OR OR involves “research on operations”. Thus OR applied to problems that concern how to conduct and co-ordinate the operations with in the organization. • OR has been applied widely in areas of manufacturing, transportation, construction, telecommunication, financial planning, health care etc etc. • Research part of OR means that OR uses an approach that resembles the way research is conducted in any established scientific fields.
- 26. Basic Terminologies Feasible solution: • Any solution LPP/ NLP which donot violate constraints are called feasible solution. • Feasible solution may be optimal (best) solution. • Any LPP/NLP can have more than one feasible solution. • Optimal solution to LPP/NLP must be a feasible solution. Infeasible Solution: • Any solution violate at least one constraint is called infeasible solution. • Any LPP/NLP have infinite number of infeasible solution. • Infeasible solution lies outside the bounded region.
- 27. Optimal solution: • It is a feasible solution that has the most favorable value of the objective function. • The most favorable mean the largest possible objective value if the objective is to maximize and smallest value if the objective is to minimize. • A LPP/ NLP can have more than one Optimal solution. Corner Point Feasible Solution: • CPF is a solution that lies at the corner of the feasible region • Every LPP with feasible solution and bounded feasible region must possess CPF solutions and at least one optimal solution. • The best CPF solution must be an optimal solution. • If LPP has exactly one one optimal solution, it must e a CPF solution.
- 28. Bounded feasible Region: • A bounded feasible region may be enclosed in a circle. • A bounded feasible region will have both a maximum value and a minimum value. Unbounded Region: • An unbounded feasible region can not be enclosed in a circle. • If the coefficients on the objective function are all positive, then an unbounded feasible region will have a minimum but no maximum.
- 29. • Linear programming is an analytical technique in which linear algebraic relationships represent a firm’s decisions, given a business objective, and resource constraints. • Objectives of business decisions frequently involve maximizing profit or minimizing costs. • Steps in application of LPP: – Identify problem as solvable by linear programming. – Formulate a mathematical model of the unstructured problem. – Solve the model. – Implementation Linear Programming: An Overview
- 30. Decision variables - mathematical symbols representing levels of activity of a firm. Objective function - a linear mathematical relationship describing an objective of the firm, in terms of decision variables - this function is to be maximized or minimized. Constraints – requirements or restrictions placed on the firm by the operating environment, stated in linear relationships of the decision variables. Parameters - numerical coefficients and constants used in the objective function and constraints. Decision Model Formulation
- 31. Summary of Model Formulation Steps Step 1 : Clearly define the decision variables Step 2 : Construct the objective function Step 3 : Formulate the constraints
- 32. Examples on LP model
- 33. For example product P consists of two subassemblies. To manufacture the first subassembly, one unit of RM1 passes through machine A for 15 minutes. The output of machine A is moved to machine C where it is processed for 10 minutes. The second subassembly starts with RM2 processed in machine B for 15 minutes. The output is taken to machine C for 5 minutes of processing. The two subassemblies are joined with a purchased part in machine D. The result is a finished unit of P. Product Q is manufactured by a similar process as indicated in the figure. The rectangle at the upper left indicates that one machine of each type is available. Each machine operates for 2400 minutes per week. OE stands for operating expenses. For this case the operating expenses, not including the raw material cost is $6000. This amount is expended regardless of amounts of P and Q produced. Our problems include the following: Find the product mix that maximizes profit.
- 34. There are many problems that might be posed regarding the PQ situation, but we choose the problem of allocating the times available on the machines to the manufacture of the two products. The decisions involve the amounts of the two products. P= amount of product p is produced Q= amount of product q is produced Maximize Z = 45*P + 60*Q - 6000 Subject to: 15*P+10*Q<=2400 (Constraint for M/C A) 15*P+30*Q<=2400 (Constraint for M/C B) 15*P+5*Q<=2400 (Constraint for M/C C) 10*P+5*Q<=2400 (Constraint for M/C D) P<=100 (Marketing Constraint) Q<=50 (Marketing Constraint ) Decision Model Formulation