Optimized Supply Chain Management
Skip Navigation Links Optimized Supply Chain Management by Joseph Lilly

 Abstract

Manufacturers of seasonal products, especially with batch configurations, have a particular challenge in managing the supply chain. Quantitative techniques, such as forecasting, aggregate planning and optimization can bring order to the chaos.

Seasonality
Consumer product markets are full of seasonal products. The list goes beyond just the snow blowers and lawnmowers, the examples we commonly associate when teaching about complementary seasonal products. I first experienced seasonality up close when working for a major U.S. poultry company in the production of holiday turkeys. In that case, it was a problem to deal with a January glut because of the hens: toms sex ratio. Although it wasn't pleasant, inventorying surplus frozen meat was the typical solution. Sometimes seasonality exists because the market has artificially created it. Even such products as consumer electronics experience a spike in demand each year during the Christmas shopping season. Management of seasonality requires a strategy for bringing products to market at the time they are demanded. There are a few alternative strategies, and each has its pros and cons.

Aggregate Planning
Aggregate Planning is a planning exercise concerned with the big picture strategy of just how to produce product quantities in order to meet (seasonal) demand. One approach is to acquire or build enough plant capacity to manufacture products "just-in-time". Yet depending on the degree of demand variability, this may require such excess capacity that much of the investment remains idle during non-peak periods. Another approach is to acquire or build just enough capacity to maintain a level-loaded operation. This requires that production during non-peak periods be inventoried and built up in anticipation of the busy season. There are other variations too, such as hiring and laying off (or adding shifts), and subcontracting. Each of these is not without issues and costs. Management must carefully consider all the issues such as how well the products inventory, how disruptive the process is of hiring new employees and starting up another shift, how available and expensive is contracting out production to other manufacturers. Upon selection of an appropriate strategy a quantitative model or aggregate plan can be designed and published within the company.

Forecasting
An aggregate plan depends on a quality forecast. A poor forecast will produce a poor plan and almost always leads to a crisis mode of management. An effort to produce a reasonable forecast usually pays off because all other plans for materials, labor, equipment and capacity, and financing will have some reasonable basis. With my clients, on the subject of forecasting, I sometimes hear responses with a tone of sarcasm, asking whether I have a crystal ball! In truth, forecasts are almost always incorrect. Yet planning that incorporates a systematic forecast, and which compensates and adjusts for a margin of error, are almost always better than no forecast. There are a number of forecasting methods available and as a young analyst, one of the first principles I learned was this: Analyze the data for each product and select an appropriate forecasting model. One exercise that has served me well over the years is known as T-S-N analysis. This is a technique in which a product's historical demand can be quantified and explained in terms of how much the variability is due to trend, how much is due to seasonality and how much is due to noise. For a product having a high degree of seasonality, a forecast that makes a seasonal adjustment will almost certainly be more accurate. Using a simple moving average forecast for a product having a strong trend will usually be lagging and so less accurate. There is a tradeoff between forecast accuracy and forecast cost. High value items justify more cost and effort in making a forecast. Forecasts for most consumer goods can be made automatically and repetitively, and except for some initial analysis, these are cheap and effective.

Make-to-stock
Some industries are subject to seasonality on both the supply side and the demand side of the equation. Forest and building products for example are subject to winter snows where logging operations especially in the western U.S., come to an abrupt halt, and a building season that usually slows during the winter. Many food products and for example wineries, can only receive their raw material supply during the autumn "crush" which only lasts for a month or two. Products with characteristics that make inventory storage practical are typically manufactured in a make-to-stock environment. Inventory can serve the important role of "decoupling" supply from demand.

Batch Manufacturing
Manufacturers of high volume and especially mature products, naturally tool their facilities for efficiency. Machines setup to repetitively process many pieces per hour, or containers that pass large quantities through a process have resulted in lower cost per unit due to economies of scale. Generally, producers of seasonal, make-to-stock products are batch manufacturers. Yet this batch strategy presents the inventory manager with some serious challenges. In a make-to-stock environment, maintaining target inventory levels is a difficult proposition. A product's inventory is routinely overstocked at the end of a large batch run, or run down to a dangerous level while waiting for the production of other product batches to be completed.

Cost of Inventory
Although inventory serves a vital role, the costs should be recognized. The item cost, whether by purchase or manufacture, represents real money that sits idle instead of earning a return, if even in a bank savings account. Storage costs include those expenses of owning or renting storage space and the equipment and labor associated with material handling and the general facility operating expenses. There are inventory costs associated with risk. There are risks that items will become damaged, lost or stolen, that they will become obsolete or will physically deteriorate. Inventory also tends to conceal quality problems in the manufacturing processes because some problems do not become apparent until discovered by the end user. In the meantime, more defective or reject production has been completed which may have to be reworked, discounted or discarded. In recognizing the nature of these costs, management should always seek to keep inventory at a minimum.

Master Production Scheduling
Master Production Scheduling implements the aggregate plan by routinely setting the daily or weekly production schedule in detail by item. The MPS is the important communication link between Marketing and Production. Often, the Master Planner is in a seemingly hopeless situation. This role requires the accommodation of demanding customers always expecting to receive their orders on time. He or she must also work with Plant Managers who are pressed to operate with long production runs and with as little changeover downtime as possible. There tends to be a conflict between competing objectives.

Typical MPS software programmatically triggers order release anytime a product’s inventory has reached its reorder point. Establishing item reorder point is one objective of the forecasting and aggregate planning work. While analyzing time series data, the degree of expected order variability can be quantified. A reorder point in part, expresses a safety stock target, which serves the purpose of covering the unexpected, above average demand occurrences. (A reorder point also expresses lead-time; the time period between order release and order receipt). Quantifying target inventory quantity is also a function of how much risk management is willing to assume: The more risk, the less safety stock, the less risk the more safety stock. With seasonal products where the aggregate plan calls for building inventory during the off-season, the target and reorder point should move from month to month.

In practice, the Master Planner must consider the company’s productive capacity and so regardless of how demand might outpace the forecast, the schedule must be feasible within that capacity. With extremely seasonal products, this can become a problem during that peak time of year when inventory draws down to depletion. The Master Planner must also observe the aggregate plan and if in the case of a level loaded strategy, must schedule a full week of production even during the off-season if sales should drop below forecast. This can be a problem when scheduling additional production will become excess beyond the targeted inventory levels.

A case for the Optimized Master Production Schedule
I like to think of the products as a list of items that compete for the plant’s limited resources. If one hour of time is scheduled to production of Product A, then that capacity is not available for Product B. That one hour is gone forever. Product A has consumed it. So there are tradeoffs between competing products. If Product A requires tooling or batch processing such that we should dedicate a full 8 hour shift to its production, then Product B still looses access to that capacity and for 8 hours. If in producing a batch of Product C, we generate so much quantity that it will take a long time for sales to draw down inventory, then we will be less inclined to schedule its production. Yet if inventory of that Product C is depleted and an important customer has placed an order, then we are compelled to schedule its production anyway. If these issues exist on a long list of products then we will start to consider producing in smaller batches. Yet if this causes much downtime for numerous line setup/retooling then that decision will cause our productive capacity to be significantly reduced and we’ll put ourselves into a crisis mode. It is just these types of issues that we can resolve with the use of optimization modeling.

It is unrealistic to think that anyone could possibly keep track of and mentally resolve all these issues by experience and intuition alone. Fortunately, there is a systematic method for capturing all this detailed information along with the interrelationships and tradeoffs and for arriving at a logical conclusion. That method is optimization modeling. Optimization, or linear programming, is simply an application of algebra. It is a way to maximize or minimize some objective such as profit or cost. The details of an operation can easily be described with simple algebraic equations (or inequalities). Optimization uses a special algorithm to solve these equations and reveal the combination of decisions that are optimum. The method considers all the tradeoffs, and determines the best program with respect to the entire operation.

In my prior supply chain work, I have modeled such inventory problems as something of “a game of penalties and rewards”. In scheduling production, I give the highest rewards to our highest priorities. I give my clients discretion on this but usually filling immediate orders of high valued items is prioritized the highest. Scheduling production of items for the purpose of bringing inventory up to the current target level receives a moderate reward and to targets 8 weeks from now receives the lowest reward. In a make-to-stock environment, we schedule production to make an inventory target usually because of a forecast. In some cases, the target is due to an actual order. Since we’re in the business of actually selling items, the hard order takes priority over the soft forecast. In these inventory problems, the decision to leave hard orders in a shortage situation is heavily penalized, and producing beyond the target inventory is moderately penalized. These inventory models are also formulated subject to the available plant capacity and batch sizes. Because of the batch size increments, the current inventory on hand, and the current order quantities, it is practically impossible to devise a schedule where every item’s inventory is exactly at target. Solving the model returns a solution of batches of items that should be scheduled in week 1, week 2, etc. The solution is “with all things considered” so that the tradeoffs have all been balanced and so the schedule is one that maximizes the rewards while minimizing the penalties.

An optimized MPS keeps the short-term weekly schedule in line with the longer-range aggregate plan. Even when orders do not come in perfectly as the forecast had predicted, optimizing against both actual orders and the moving inventory targets, the system tends to be self-adjusting. This method works equally well during either the high or low seasons. More importantly it has proven to be effective in both reducing inventory costs and also improving customer service. One client reported that with the optimized MPS, they were able to reduce their inventory from $16 million down to $12 million and also reduce order lead-time from 4 weeks to 3 weeks.

Joseph Lilly is the President of JM Lilly and has been a consultant for over years. His business to apply quantitative methods to solve industrial problems and in particular, to help clients optimize their resources.
Copyright ©2009 JM Lilly  All rights reserved worldwide.