Supply Chain Frontiers issue #21. Read all articles in this issue
What is the most efficient blend of make-to-order and make-to-stock manufacturing that also meets a company’s profitability goals and satisfies the ever-increasing demand for a wider choice of products supplied at the lowest cost and quickest delivery time? It is a conundrum that many manufacturers are facing. A possible solution has been developed by Shardul Phadnis, a graduate of CTL’s Master of Engineering in Logistics (MLOG) Class of 2007.
In make-to-order manufacturing companies produce items based on actual customer orders. The method requires little or no finished goods inventory, which means that carrying costs and product obsolescence rates are low. On the other hand, demand fluctuations can cause extreme swings in capacity utilization, setup costs can be high, and lead times are relatively long, because orders are not filled from readily available inventory. Make-to-stock items are produced to build a finished goods inventory which is used to fulfill demand. That makes production planning easier and more cost-effective, and enables the manufacturer to offer short lead times. On the downside, this method relies on demand forecasts which are notoriously unreliable.
A hybrid manufacturing approach that plays to the strengths of each option offers considerable advantages. The challenge is how to determine which items are best suited to each method; make-to-stock favors items with regular demand, while make-to-order is more appropriate when demand in sporadic.
One way to create production plans for the make-to-stock items is to calculate the economic production quantity and period for each item that minimize the total setup and inventory holding cost, and then determine safety stock level using a stochastic-demand model (where demand can be determined, but there is also uncertainty in how it may vary). But such methods have limitations. For example, one approach uses the so-called “min-max” policy (the minimum and maximum levels of stock required, where minimum can be the point at which reordering is triggered, and maximum can be expressed as delivery quantity plus safety stock) to schedule production of multiple items. The problem is that this approach can lead to situations where more than a single item needs to be produced simultaneously, when it is only possible to make one at a time.
Planning for make-to-order production also involves some thorny problems. Production can be planned only when an actual order is received. It follows that the manufacturer needs to decide how much capacity to reserve in each production period and what lead time should be quoted for each item - a difficult challenge.
In addition, the production plans for make-to-stock and make-to-order items need to be combined to form a cohesive schedule that provides the desired level of customer service at the lowest cost without violating any capacity and technology constraints.
These issues were taken into account in the development of new production planning policies for a chemical manufacturer. The company produces several variations of an interlayer film used in automotive glass. Each SKU is differentiated by six characteristics of the film: adhesive, color, width, length, width of colored band, and roll orientation. All variations are produced on a continuous-production line that can make at most two SKU’s with the same adhesive and color at a time.
These complexities incur high changeover and inventory costs, and the production of multiple items requires careful planning. The goal was to develop a production planning method that guarantees at least 95% availability of the finished goods at the lowest cost.
In this two-part solution a heuristic simplifies the problem and refines possible remedies. For example, it ensures that production plans developed for individual items are feasible in a multi-item environment; and it ascertains what base stock is needed for each item by taking required safety stock and optimal production quantities into account. The second part of the solution applies the heuristic to the manufacturer’s production planning policies. Product groups for the different SKU’s manufactured are created in order to rationalize switching costs.
Simulations of the new production plan showed that average material availability was 98.3% and ranged between 95% and 100% for individual items. But this plan pertains to product groups; the next step is to generate an SKU schedule from it. To accomplish this, SKU’s are segmented into make-to-stock and make-to-order categories. A regular schedule is developed for the make-to-stock items. Product groups are formed for the make-to-order items and, instead of scheduling their production, only capacity is reserved for each group. When actual demand for a make-to-order item is known, it is produced during the period in which capacity is reserved for the product group to which it belongs.
The work presents a heuristic for scheduling a complex production system, where inventory planning and production scheduling problems need to be solved together. The cost of the solution is within six percent of the optimal cost. Moreover, simulations indicate that the production plans meet the manufacturer’s capacity constraints and customer service requirements.
This article is based on Shardul Phadnis’ 2007 MLOG thesis "Inventory Segmentation and Production Planning for Chemical Manufacturing" that was awarded the Outstanding MLOG Thesis prize for 2007. MIT Professor Steve Graves was the thesis advisor. For more information on MLOG theses contact Chris Caplice, MLOG Executive Director. Also, see the news item Get Your 2007 MLOG Research Journal.