A case study: reducing the cost per order and improving productivity within a chemicals distribution center.

This is the first of two articles that provide an in-depth discussion of how a major manufacturer and distributor of chemicals recently used metric and process benchmarking to reduce its cost per order and improve the efficiency of its main distribution center. This first article describes how the company used benchmarking to identify problem areas and develop effective solutions to these problems. The second article describes the solutions that were developed, the implementation process that was used, and the results that were achieved.

THE CURRENT SITUATION

In mid 2003, a manufacturer and distributor of specialty chemicals for the construction industry found that it was receiving an unacceptable and increasing number of complaints from its customers. The complaints were focused primarily on quality and service issues related to distribution. These complaints included missed delivery dates, incomplete orders, and missing products. In addition, the company's costs per order were rising.

For years, the company's distribution centers (DCs) had used a combination of computers and paper systems. When products were received from the manufacturing sites, they were placed in any floor location that was available at the time. The locations were recorded on paper and later entered into a computer locator system. There were numerous recording errors and ongoing delays in product shipment due to the lag time between putting the products away and entering the data in the locator system.

To fulfill customer orders, a paper "pick ticket" was printed and then used by warehouse personnel for picking product orders. The individual line items were listed on the pick ticket in the order that they were entered into the computer by customer service. This was clearly not the most efficient picking order. This caused excessive and wasted time traveling from pick location to pick location. The pick tickets were generated using a computer locator system that recorded the location of the product but not the quantity of product that was in the particular pick location. This caused the same inventory to be allocated to multiple orders.

One solution proposed by the company's information systems (IS) group was to implement a warehouse management system to help control the warehouse processes. While a computerized warehouse management system could help solve some of the problems that had been identified, the Director of Distribution was concerned that there were more basic problems that had not yet been identified. The Director of Distribution was determined to understand why the distribution department was unable to meet customers' expectations and to determine if a warehouse management system could help reduce customer complaints.

To accomplish these objectives, he authorized a study of the processes and procedures in use at the six distribution centers. In addition to understanding how the distribution centers operated, it was decided that the company's six DCs should be benchmarked against each other as well as against other distribution centers within the same industry. Since it seemed unlikely that competitors would be willing to share sensitive metric and process benchmark data with the company, it was decided to use an outside consulting company with industry experience to collect and analyze the data. This would ensure that for each metric the calculations were the same and allow for direct comparisons. Prior to this benchmarking program, the company did not have standard metrics for measuring distribution center performance other than total dollars shipped.

BENCHMARKING THE WAREHOUSES

It was decided that the benchmarking program would include both metric and process benchmarks. The metric benchmarking would provide measurable evidence of how well the company DCs performed when compared to the best in class distribution centers. The process benchmarking would show why the company DCs were achieving those results.

The metric benchmarks concentrated on costs and productivity. The process benchmarks focused on order fulfillment processes, safety, and service. The corporate IS department was tasked with providing the data needed for the metric benchmarks. When the data was provided, it was summarized and compared to previously published results as a check of its accuracy. For example, the total pounds shipped per month for each DC was compared with the results published in monthly reports. It took several attempts before it was agreed that the raw data was the same as the data included in the published results. The data for calculating the metric benchmarks for outside companies was collected from previous benchmarking projects at the outside distribution centers.

Two-day on-site visits were scheduled at each of the company's DCs to observe and record the processes and procedures that were in use in day-to-day operations. During these site visits, a number of tasks were completed including reviewing the order fulfillment process documentation, observing and documenting all key processes in the DC, understanding building and equipment limitations, and having discussions with the DC managers.

Metric Benchmark Results

To the surprise of the DC managers, the company DCs were far behind their competitors in two key metric benchmarks: cost and productivity. When the average cost per order was compared to the industry average, the company DCs were over twice the industry average and productivity was 26% lower than the competition. With one exception, none of the company distribution centers stood out as being significantly better in costs or productivity than the others.

Figure 1 shows the relative cost per order for each of the distribution centers when compared to the industry average. For example, DC 1 has costs per order that are 2.7 times the industry average costs. The cost per order ranges from 1.1 to 11.6 times the industry average.

Figure 2 shows the relative productivity as measured by lines per person hour when compared with the industry average. For example, DC 5 is running at 58% of the average industry productivity. Productivity at the DCs ranged from 23 to 157% of the industry average. DC 4 was the only DC to beat the industry average.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

Since none of the company distribution centers was significantly better than the others and all of them performed poorly when compared to the industry average, it led us to believe that there was something in our buildings, equipment, or processes that was causing the DCs to be outperformed by the competition.

Since none of the company DCs performed significantly better than any other and they all performed poorly compared to the industry average, it was concluded that there must be something in the buildings, layout, or processes that was causing the poor results. There are two reasons why the results for DC 2 are so poor. The shipments from this DC are sporadic because this DC supports a new export business. They have not reached a volume of shipment to allow them to fully utilize the warehouse personnel. Secondly, the majority of the product is shipped by sea. This means orders are often held until a full container can be shipped.

Process Benchmark Results

Brown paper flow charts of the order fulfillment processes in use at each of the DCs were compared. The processes that were studied included receiving, put away, order picking, and shipping. As might be expected, each DC used similar processes to complete the above tasks. There were minor differences between the DCs to account for building or equipment differences but not enough to explain the results of the metric benchmarking. None of the DCs had any processes with built-in inefficiencies such as requiring an operator to move three or four pallets to ship a particular batch of product. It was concluded that the processes used to fulfill orders did not explain why the DCs performed poorly.

[FIGURE 3 OMITTED]

When studying the metric benchmarks, it was noted that another of the benchmarks showed poor performance. That benchmark was storage utilization. Storage utilization is a measure of how much of the cubic volume of the building is being used. The average storage utilization at the company was 34% compared with the industry average of 80%. During the on-site visits, it was noted that the DCs used racking sparingly. Most product was stored in floor locations. In effect, the DCs were only using the bottom third of the building. This meant that the inventory was spread over a much larger footprint and this increased travel time for picking and put away of products.

To check the effect that poor space utilization was having on picking, mapping was done on a sample of the paths traveled by order pickers when picking orders. Figure 3 shows an example of one of these maps. A detailed study of the map shows that when picking this order, the order picker covered nearly the entire footprint of the distribution center and had to return to one location a second time to complete the order.

In one warehouse, travel time accounted for 90% of the total time to pick an order compared with 60% in the comparison DCs. In addition to having a very large footprint, it was noted the products that were picked frequently were not located close to the shipping area, thus requiring additional travel time when picking frequently ordered products.

DEVELOPING A SOLUTION

To measure the success of the program, goals would need to be established. The goals that were agreed on were:

* Improve picking efficiency to five lines per man hour (slightly better than the industry average);

* Reduce cost per order to $40.00 or industry average;

* Eliminate non-value added travel time; and

* Reduce the physical size of the warehouses needed and sublet the excess space.

The metric and process benchmarking has shown that the DCs footprint and poor product placement were major contributors to non-value added travel time. A computer simulation was used to determine the effects on travel time of re-laying out the warehouse and placing products that were frequently picked near the shipping area. In the computer simulations, the same orders in the mapping analysis were used to provide an estimate of the improvement that could be achieved. The results of this simulation showed that travel time could be reduced by up to 50%. Additional analysis showed that the footprint could be further reduced and productivity enhanced if slower moving stock were stored in racks. The computer simulations showed that it was possible to achieve the goals that were agreed upon by changing the layout of the ware-house and by optimizing the picking paths used by the pickers.

To test this strategy in a real world situation, it was decided to implement the strategy in one of the DCs as a pilot project. Once successful, it would be rolled out to the other distribution facilities. One DC was selected and a team consisting of the DC manager, first line supervisor, operators, and an outside consultant was formed. The goal of the team was to develop a ware-house layout that would allow the company to meet the agreed goals and to implement a warehouse management system.

SUMMARY OF THE ANALYSIS

The primary causes for lower productivity were very large DC footprints, use of low-density storage, and poor placement of products within the distribution center. The pick ticket and locator systems in use were contributing to wasted time and inaccurate order picking. To achieve best in class performance, it was necessary to re-lay out the warehouses, change storage methods, reduce the footprint, change the product placement strategy, and incorporate a warehouse management system to reduce costs and increase order accuracy.

In the second part of this article, the successes and the problems encountered during implementation, and the problem solutions are discussed. The effects the changes had on cost, productivity, quality, and service are summarized.

by Kevin Reid and Don James 

Orr & Boss, Inc.*

*44450 Pinetree Dr., 2nd FL., Plymouth, MI 48170.