OPERATING AT THE PUSH/PULL BOUNDARY
The rubber meets the road where the push planning process interfaces to the pull production process. The supply chain trading partners should decide the optimum place for this to occur. When the push/pull boundary is upstream with the supplier, the supply chain is being operating as build-to-order, but the customer delivery time is too long. See Figure 2. The customer has to wait for too many operations to be completed. When the push/pull boundary is downstream with the distributor, the supply chain is being operated as build-to-stock, but the customer service level is too low. The customer experiences the high probability of a stockout depending on actual product mix. When the push/pull boundary is in the middle of the supply chain, in the factory,
neither customer delivery time nor customer service level are optimal. But only one trading partner can operate as the push/pull boundary. The conversion of push planning into pull production is a necessary but independent action from synchronization.
One optimizing solution is to introduce postponement into supply chain distribution. Postponement is used to trade information for inventory. The product is kept generic as close as possible to the customer; this may require some product reengineer-ing. A postponement center takes each customer order, and quickly converts a generic subassembly into a customer-specified final product. The postponement center is the push/pull boundary. Generic subassemblies are pushed into the postponement center, while customer specified complete products are pulled out of the postponement center. This is very effective for multitiered distribution.
A different optimizing solution is to operate the factory in short sequences of mixed model production. The push plan is to run a sequenced schedule of specific model quantities. The factory's material handling times, machine setups, and tooling change-overs have to be continuously improved to the point where mixed model production is practical. Preventive maintenance programs and trained machine technicians must be in place to deal immediately with the wear and tear of changeups. The pull production is modifying the planned schedule to actual demand consumption. Sequences are rearranged and quantities are repeated to match actual incoming orders. Mixed model production can be very effective in a supply chain where the factory ships directly to the end customer or ships through a single tier of distribution.
HOW TO DETERMINE THE SYSTEM CONSTRAINT
Eli Goldratt's theory of constraints describes how the constraint within a group of work centers limits the throughput of the whole factory (see Figure 3). The constraint produces product at its full capacity, while every other work center effectively has some excess capacity. The first work center in the chain is tied, or synchronized, to the constraint in order to eliminate the buildup of unnecessary work-in-process ahead of the constraint. The constraint is scheduled to only work on jobs that have customer orders behind them to avoid the buildup of unnecessary inventory after the constraint. And, if for some reason one of the other work centers falls behind, it can use its excess capacity to
catch up. This model has direct applicability to a supply chain where the work centers become the trading partners, the constraint becomes the system constraint, and the synchronization signal becomes the broadcast signal.1 The system constraint controls the end-to-end throughput of the supply chain. Since some of the trading partners operate with excess capacity, proper synchronization throughout the supply chain prevents the buildup of excess inventory upstream and downstream from the system constraint. For synchronization to work properly every node in the supply chain must be "capable." This means that there is adequate capacity at every node in the supply chain, yet one node will be the system constraint.
The first step is to define the capacity required per node for the supply chain to be capable, and the second step is to identify which node is the system constraint:
• Compile a list of SKUs delivered through the supply chain.
• Verify that each SKU is processed through each trading partner by comparing supply chain routings.
INVENTORY MANAGEMENT UNDER SYNCHRONIZED OPERATIONS
Under synchronized operations, there will be some inventory at every node of the supply chain and some inventory within every pipeline. When customer demand is light, node inventory will be large and pipeline inventory will be small. When customer demand is heavy, node inventory will be small and pipeline inventory will be large. In fact, the sum of the total inventory in equivalent SKU units of all the node inventory plus all the connecting pipeline inventory will be constant. If it is at all possible to eliminate some nodes in the current supply chain, then do this optimization before starting synchronized operations. This is because the total supply chain inventory investment is driven mostly by the number of nodes.
Then, before synchronized operations can begin, enough inventory must be put in place at each node to support one synchronization cycle at maximum throughput. A synchronization cycle is the time between the broadcast of one demand signal from the system constraint until the next; this is typically one day. From then on each node must manage its inventory on a first-in, first-out (FIFO) basis to ensure stock rotation.