|
Lean supply chain and two of its major components, lean manufacturing and lean distribution, focus on satisfying customer demand efficiently at the lowest cost – minimizing waste. In a distribution center (DC) this generally requires managing thousands of SKU’s simultaneously, achieving a balance that considers many variables characterizing the importance, constraints, and behavior associated with each SKU and its interaction with all other SKU’s in the facility. The interaction among SKU’s is dynamic and often difficult to predict, particularly as conditions change. Appreciating these dynamics and how to deal with them is easier to do in the context of a familiar model – your home refrigerator.
Most of us are not particularly focused on optimizing the contents of our refrigerators. Open the door and quickly find what you want when you want it – cold, fresh, and ready for pack/ship to the table for consumption. Many picks are for single SKU orders; others are items on a recipe paper pick list (RF is not yet available for the fridge). In this “PULL” model, demand may not be satisfied at the 99% level … only four ounces of milk left in a gallon jug, the cheddar has green spots, the scallions are hiding under the lettuce and tomatoes, twelve containers of yogurt (which you don’t like anyway) occupy half of the top shelf, the door bins are so full, not a single additional condiment will fit, and beer inventory is zero, resulting in a critical stock out.
Demand, supply, and storage media optimization are not in balance for the SKU’s in this forward pick module; process improvements are required! Achieving efficiency and balance – a “lean” fridge, is not too conceptually different from managing a pick module in a DC. Demand, replenishment strategy, material flow, critical items, volumetrics, safety stock, storage media, bin sizes, slotting, dead stock, reserves, returns, customer satisfaction, supplier management, and cost have parallels in the kitchen and in a modern DC.
The Fridge Pick Module Subzone:
Consider my Fridge Pick Module (FPM), a 22.6 cu.ft. subzone of the Kitchen Zone. The storage media consists of 11 static shelves and 8 bins of different dimensions currently supporting about 100 refrigerated SKU’s and 50 frozen SKU’s. Inventory turns range from one to over 100. Space, especially in the golden zone, is always at a premium, overstressed during holiday seasons. Demand spikes occur randomly with the arrival of kids home from college.
Key Variables Driving FPM Management:
To avoid a meltdown in customer sat, the following variables are considered and integrated to manage the FPM:
Demand/Forecast: The FPM is managed largely using a “PULL” strategy. (Forecast driven “PUSH” approaches have been discarded since they generated excessive inventory balances, overwhelming storage capacity and leading to excessive dead stock.) While average demand for many items is nearly constant with a small standard deviation, demand for critical items, e.g., milk and eggs, is monitored continuously to avoid stock outs. Bin reorder points (ROP’s) are established based primarily on consumption (forecast data is included to accommodate aforementioned peaks in demand). Peak inventory balances for critical items are usually accompanied by a space crunch forcing a reduction in the total number of SKU’s supported by the FPM storage media, resulting in short picks and customer complaints. While using economic order quantity (EOQ) in purchasing decisions is a cost consideration, it often creates excess inventory leading to dead stock.
Replenishment Strategy & Material Flow: Using FPM reorder points, SKU’s are replenished from reserves, (the garage freezer or the pantry) or directly from receiving (trunk of the car). While the latter is the preferred strategy for many items, particularly bulky and fast moving ones, LTL transportation and material handling costs can be excessive. The objective is to keep the FPM bins from falling so far below the ROP that safety stock quantity is inadequate to handle demand while awaiting bin replenishment (otherwise known as satisfying demand within lead time). In general, the desired flow path is directly from receiving to the FPM. This requires a delicate balance between SKU purchase order quantities tuned to ROP’s, reserves capacity, and lead time.
Critical items: Using a Pareto analysis, SKU’s that are expensive and/or large and/or have long lead times and/or have limited shelf life are at the top of the food chain in terms of management attention. Spending analysis energy on a cheap, small, low demand, short lead time item like a lemon is less important than milk or beef – cheese is probably somewhere in between.
Volumetrics: The size of a SKU stored in the FPM is a key variable since space is limited. Knowing the dimensions of an item is critical for anticipating which storage media and how much of it can be allocated to satisfy demand. Here, package quantity, EOQ, and replenishment lead time also come into play. With ample reserve storage, buying a case of Perrier on sale may be better than two liters at regular price only if demand dictates.
Safety Stock: Safety stock levels should be set to accommodate fluctuations in demand and the vagaries of supplier/transportation performance. In the best case, this level is set for each SKU, but needs most attention for Pareto class ‘A’ items. Fill rate/customer satisfaction can be traded for lower safety stock levels as a cost/space consideration. For short lead time items, a quick trip to the store, while inconvenient, may be necessary.
Storage Media and Bin Sizes: With limited space in the FPM, deciding how much space to allocate to a given SKU requires some math. All of the variables discussed so far come in to play, with demand being the most important. The first step is to calculate the optimal bin quantity based on demand within lead time plus safety stock. This yields the desired quantity of an item, independent of its size. Using this value, multiply by the item cube to determine the optimal space required for each SKU. In a best case scenario, the bin size will be twice the optimal bin quantity space plus room for safety stock – a classic multi-bin Kanban. Milk is a good example; a new gallon needs to be in place before the current one is empty. Intuition suggests that this rule allocates too much space for some SKUs. Be careful, length-width-height dimensions will also limit your slotting choices. Don’t forget to account for space between and above items to allow for effective picking. Depending on lead time and criticality, less space for a SKU may be allocated at the risk of a stock out. Packaging options, e.g., half gallons of milk, may ease pressure on storage media.
Slotting: In the FPM Subzone, slotting is based on the available media, demand, cube, and proximity to the fridge door. While this analogy to a DC is limited, there are similarities. The key issue is the tradeoff between cube and popularity. Many small fast moving items can be stored in the same space as a single larger fast mover; which deserves slotting priority? Back in 1963, J.L. Heskett of Ohio State University1 introduced a concept known as Cube Per Order Index (CPOI). Simply stated, CPOI balances the space required by an item (bin size) with the number of times it gets picked in a given time period – bin cube/picks. The lower the CPOI, the closer it should be slotted to the pack/ship area. Using this approach, the canned dog food using 45 cu.in., picked twice a day (CPOI = 22.5) deserves preferential slotting over the english muffins requiring 150 cu. in. picked once a day (CPOI = 150). Both compete for space on the same shelf, but the dog food requires less space per pick. (Note that popularity and demand, while related, are not the same, but can be both be included in more advanced CPOI calculations.)
Dead Stock: A major obstacle to the lean fridge is the collection of items that never seem to get eaten. They occupy valuable space and generate unnecessary cost, e.g., digging through the vegetable bin only to realize that many items should be moved to the garbage bin or donated to the local bird population. Eliminating dead stock requires discipline – periodic audits and a proactive approach to excessive stocking due to overestimation of demand and aggressive purchasing of items on sale (EOQ).
Reserves: Replenishments for the FPM can come directly from receiving or from reserve storage subzones, the garage freezer and the pantry. While putaways to the FPM from receiving (the car trunk) simplifies material handling and therefore reduces costs, the lead time is longer than a replenishment from reserves; it takes only a minute or two to resupply the grape jelly. Like the FPM, reserves have finite capacity and need to be managed in a similar fashion. Because the reserves have a larger capacity than the FPM, they are prone to mismanagement since looming problems can go unnoticed. Accumulation of dead stock and slow movers represent waste and put pressure on storage media. Poor inventory visibility or slotting exacerbates the problem since items may become “lost” prompting unnecessary purchase orders (and attendant costs).
Returns: Leftovers, like returns to a DC, present challenges for the reverse logistics process. Some are not worth saving and should be written off immediately. The remainder may require special packaging that is often inconsistent with available media and represent a new SKU for which adequate bin space has not been allocated – putting pressure on other SKU’s in the limited FPM environment. Storage in reserves (garage freezer) is an option, but often results in dead stock which ultimately gets discarded anyway. The value, volume, packaging, and arrival rate of returns are important considerations. Pareto class ‘A’ and ‘B’ items in original packaging may be easily restocked and worth the effort. Since exception processing is much more expensive than standard DC processing, alternatives like outsourcing to a 3PL for liquidation may be cost effective.
Customer Satisfaction: The key indicators are variety, quality, and fill rate. The variety and quantity of SKU’s stocked must be in balance with storage space and media type constraints. Demand, particularly for critical items, should modulate the variety of items supported in the FPM. Low demand items (anchovies) may be stored in and picked from reserves. Quality is linked to freshness (inventory turns). And fill rate is a function of proper bin quantities to satisfy demand within lead time. A good, well executed replenishment strategy addresses all of these issues. (The replenishment strategy is often the weakest link in DC processes!)
Supplier Management: Aggressive comparison shopping is analogous to a good purchasing organization. Paying attention to demand, shelf life labels, packaging consistent with storage media, and lead time can all reduce waste and maintain high customer service levels. Using a limited number of strategically located suppliers (stores) that offer a wide selection also minimizes transportation and material handling costs.
Cost: In managing the FPM, the price of a given SKU is not included in operational cost, although waste, e.g., dead stock, will certainly affect the overall household bottom line. Carrying costs are negligible. The operational cost focus should be on ordering, transportation, and material handling costs. In this model, ordering translates into managing purchase orders, i.e., what, how much, and when to buy. Transportation is the to/from store travel cost including labor. Material handling covers labor for picks, putaways from receiving, replenishment, and stock maintenance in the FPM and its reserves. Taken individually for this “DC” the costs may not seem like much – unless you have to pay someone even minimum wage and transportation at $0.45/mi. to do the work for you.
The Bigger Picture:
For the Fridge Pick Module, managing the interaction of this small number of variables can be done manually, given some attention (and the willingness to throw things out). By comparison, the commercial DC is nearly as close in the supply chain to the customer and can be managed with a similar “pull” strategy. In contrast, it is a much larger challenge. The complexity associated with a strategy that considers even the limited number of variables discussed here is daunting for 20,000 SKU’s. Unfortunately, in many DC’s this is a back-of-the-envelope exercise. While much of the data required to support a “get and stay lean” program in a modern DC can be easily derived from purchasing, RF warehouse management, and transportation systems, modeling the DC processes to take advantage of the data is exponentially more difficult. Focusing on individual aspects of the problem, e.g., demand or slotting, does not cut the dynamics mustard to achieve a lean DC environment. Implementation requires commitment to an analysis strategy and the tools to integrate any and all of the data available.
This refrigerator analogy is intended to stimulate thinking about the complex relationships among a relatively small number of items in a familiar context. The tools and techniques to achieve a large-scale lean DC with more variables and a full complement of SKU’s will be explored and demonstrated in follow-up articles stressing
lean inventory by design.
1 – Heskett, J.L. 1963 “Cube-Per-Order Index – A Key to Warehouse Stock Location”, Transportation & Distribution Management 3, pp27-31
Contact Marc Barnett at: mbarnett@chitechinc.com
Telephone: (732) 332-1012 X 10
|
Introduction
