Departemen Teknik Industri DYNAMIC SYSTEMS Lecture-08 Departemen Teknik Industri Static Systems • Decision processes have been studied (Static Systems) • Concerned with making a single decision at a given point in time (usually at present) • For the case of a periodic decision –it is considered that the future is stationary, hence each decision is identical to the first one. Departemen Teknik Industri Dynamic Systems • The time element is explicitly incorporated into • • • • • the decision process Problems studied involve a sequence of decision points over time The future is not stationary (but dynamic) Hence, the state of the system changes over time Each future decision point may face a different state of the system Hence, consecutive decision are not identical Departemen Teknik Industri Dynamic Systems • How long into the future we should look? • How long is the planning horizon? • Planning horizon is the length of time that the model covers • For dynamic situation, the choice of the planning horizon becomes more critical • The sequence of decisions is called the decision policy • Objective: Finding the best sequence of decisions Departemen Teknik Industri Planning Horizon • Number of weeks, months, or years covered by the model • What is the most appropriate length for dynamic models? The longer the costlier data collection and the computation of the optimal policy Keep as short as possible • Systematic sensitivity analysis with regard to the time interval covered may help in assessing the effects of future events on the initial set of decisions Departemen Teknik Industri Planning Horizon • Aim of Modeling Dynamic Situations To determine the best set of decisions for which a firm commitment has to be made at the beginning of the planning horizon Departemen Teknik Industri Planning Horizon for Seasonal Process • The nature of the problem may strongly suggest • • a suitable length for planning horizon This is the case for decision problems subject to a seasonal pattern for some of their inputs. For example: A hydro reservoir (p.359) The planning horizon is one year A natural break between consecutive seasonal cycles (at the end of the season the reservoir is empty) Carry over from one cycle to the next Create an artificial break between consecutive cycle (p.360) Departemen Teknik Industri ROLLING PLANNING HORIZON • For the case with no natural breaks or end points, a rolling planning horizon approach is used. • A planning horizon T periods is used to determine the initial set of optimal decisions • Then, the decisions are implemented and their effects state of the system recorded • The problem is again analyzed based on the update state of the system (See Fig.14-1) Departemen Teknik Industri Departemen Teknik Industri PLANNING HORIZON LEAD UP TIME • • • • • • Implementation of the decision may not be instantaneous –takes several days, weeks, or months Need to be planned prior to the planning horizon (a lead-up time) Also, time needed for developing the overall plan Hence, this development time is also part of leadup time A lead-up time becomes part of the effective planning horizon The effective planning horizon consists of two parts - the lead-up time and the active planning horizon Departemen Teknik Industri CASE: CRYSTAL SPRINGS MINERAL WATER COMPANY • Situation Summary • Choice of Planning Horizon • Influence Diagram • Spread Sheet Mathematical Model • Finding The Optimal Production Plan • Considerations for Practical Implementation Departemen Teknik Industri Departemen Teknik Industri Departemen Teknik Industri Departemen Teknik Industri Departemen Teknik Industri Departemen Teknik Industri Departemen Teknik Industri Departemen Teknik Industri