Question 2 of 8: Materials Handling, Inventory Control and Forecast-Model Selection

Question 2: Materials Handling, Inventory Control and Forecast-Model Selection (20 marks)

Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.

Part (a) — objectives of materials handling. Materials handling adds cost but no form utility: a part is worth no more after it has been moved than before, so the governing objective of a handling study is to move material as little as possible, and to move it as cheaply, safely and reliably as possible when movement is unavoidable. Within that principle a Canadian plant engineer normally writes the objectives as follows. First, reduce handling cost, which in discrete-parts manufacturing typically absorbs a quarter to a half of total manufacturing cost once labour, equipment, damage and space are counted. Second, reduce work-in-process and throughput time: material that is being moved or is waiting to be moved is inventory, and every extra move lengthens the manufacturing lead time on which the master schedule depends. Third, improve safety and working conditions — manual handling is the single largest source of lost-time musculoskeletal injury reported to provincial workers' compensation boards, so mechanizing lifts and eliminating awkward postures discharges a duty under the provincial occupational health and safety statute as well as saving money. Fourth, improve space utilization, both floor area and cube, by using overhead conveyors, stacking frames and narrow-aisle trucks. Fifth, protect the product: handling damage, contamination and mixed lots are quality failures caused entirely by movement. Sixth, increase productive capacity by keeping machines fed, which is what an automated storage and retrieval system or an automatic guided vehicle fleet is actually bought for. Finally, support control: a handling system that also captures identification at each move (bar code, radio frequency identification tag) is what makes shop-floor control and inventory record accuracy possible.

Part (b) — what an effective inventory control system must accomplish, and the vital areas. An inventory control system exists to answer three operational questions repeatedly and cheaply: what to order, how much to order, and when to order it, at a total cost of ordering, holding and shortage that is as low as the required service level allows. Concretely, an effective system must maintain record accuracy high enough to be trusted without counting (cycle counting rather than an annual physical inventory is the usual mechanism); it must provide the planned level of customer service, stated as a fill rate or a probability of no stock-out during lead time, rather than an unquantified promise; it must keep the investment in stock within the financial limits set by management, and report that investment by class; it must give early warning of obsolescence and of slow-moving items; and it must connect to purchasing and production planning so that a replenishment signal actually causes a purchase order or a shop order. The vital areas to consider when designing such a system are: classification of items by annual dollar usage (ABC analysis), so that control effort is spent where the money is; demand characterization — independent demand items are controlled by reorder point or periodic review, dependent demand items by material requirements planning, and confusing the two is the classic design error; the cost structure, that is realistic ordering, set-up, carrying and shortage costs; lead times and their variability, which size the safety stock; the record and transaction system, including receipt, issue and scrap transactions and the accuracy audits that police them; the review policy, continuous or periodic; and organizational responsibility, meaning a named owner for each decision rule and a management report that exposes exceptions rather than burying them.

Part (c) — factors influencing the choice of a forecasting model. The choice is a matching problem, not a search for the most sophisticated technique. The factors that decide it are: the time horizon (short-term operational forecasts favour exponential smoothing and other extrapolative methods, while long-term capacity and facility forecasts require causal or qualitative methods); the pattern of the data, that is whether the series shows level, trend, seasonality, cycle or randomness, since a model without a trend term cannot track a trending series; the quantity and quality of historical data available, which is decisive for a new product where no history exists and judgemental methods such as the Delphi technique or analogy must be used; the required accuracy and the cost of a forecast error, which is what justifies spending money on a better model; the cost of the forecast itself, in data collection, software and analyst time; the number of items to be forecast and the frequency of updating — a stock list of fifty thousand items forecast weekly rules out any method needing analyst judgement per item; the availability of independent variables that genuinely lead the series, which is the precondition for a regression or econometric model; the lead time available before the decision must be made; and the ease with which the method can be explained to the managers who must act on it, since a forecast that is not believed is not used.