AAP St. Mary's Material Handling Case Study

Key Technologies

• 15 aisles of Daifuku AS/RS buffers providing about 15,000 storage locations for WIP wheels, controls systems, and inventory systems
• 3700 feet of chain driven 24v Motorized Drive Roller
• BMH Controls Warehouse Control System (WCS)
• ASAP Automation Warehouse Management System and AS/RS Controller
• Cognex/Integro vision systems
• Blue Arc Engineering custom lifts

Business Challenges

• AAP was searching for a method to streamline and synchronize the flow of wheels between various operations
• AAP wanted to reduce employee turnover and associated training costs, and lessen the manual labor requirements for handling each wheel
• Forklift traffic was dangerous and needed to be reduce or eliminated
• Visibility of inventory and system capacity needed to be improved. Scheduling and inventory locators existed as hand-written notations on loose paper

Results

• AAP St. Mary’s manual labor requirements were significantly decreased
• Fork lift traffic was safely reduced
• Inventory and quality control improved
• The project’s success has greatly improved AAP’s ability to succeed in a very competitive industry
• BMH has been asked to partner with AAP on future growth phases

Case Study Video- Script

AAP was founded in St. Mary's, OH in 1988. A division of Hitachi Metals, Ltd., they were primarily a supplier to Honda, but have added Ford, Subaru, and to a lesser level Nissan and Toyota to their customer base. Their niche in the industry is that they can produce large production runs of a variety of wheels, but can also turn around new designs very quickly that are often produced in small batches. This ability to adapt to new products has made them the preferred development partner to these accounts.

If you look at cast aluminum wheels long enough, you realize that they are probably more standard now that the old stamped steel wheels that we think come with base model automobile. AAP produces wheels from 15" to 22" diameter. They produce what is refered to as a shiny wheel, as well as a fully painted wheel. Shiney wheels are partially painted, and then machined to expose "shiny" aluminum. All wheels are clear coated for protection.

AAP is located in St.Marys, OH along US Highway 33 and State Road 66. Interstate 75 is 10 miles to the east down Highway 33. The city is on the east side of the Grand Lake of St. Marys which was man made to be the "reservoir" for the Wabash and Erie Canal.

The AAP system consists of a large, mezzanine installed conveyor network using about 3700 ft of zero pressure accumulation conveyor, 15 aisles of AS/RS buffer providing about 15,000 storage locations for WIP wheels, control systems and inventory systems that manage the process. The primary purpose of this system is to move wheels efficiently between casting, paint and machining with no human contact with the wheel. Phase 1 of the project installed about half of the conveyor and the western buffer. Phase II will launch this spring, and will complete the project.

Prior to this project AAP had just invested $25,000,000 in a building expansion that did not improve productivity. AAP was under pressure to find ways to reduce overall costs. AAP's controller took the lead to investigate possible solutions. The main goals were to reduce turnover and associated training cost, reduce manor labor going onto every wheel.

As the process was studied, fork truck traffic became a very significant issue, as did the physical activity of accessing needed WIP that was stored Ad Hoc throughout the building. There was frequent rescheduling of resources caused by the inability to efficiently access – and in some cases – even locate the WIP needed. Scheduling and inventory locators literally existed as hand written notations on loose paper.

The primary design challenge was getting the customer to think through and agree upon a disciplined, well thought out process that could then be designed around. Automating the way they were doing things before this project would have only resulted in faster chaos.

Production control, for example, was largely a mysterious process understood and operated by a few individuals with no systemic oversight. It was reactionary, and depended on their ability to find WIP wheels that had been stored in whatever nook or cranny – or offsite warehouse – they could find. Often a scheduling requirement would require the movement of many floor stacked wheels to get to the ones needed for a critical shipment.

All wheels were moved from one process to storage and then to the next process by fork truck. All of the process steps occurred at or near the South end of the building, while the only storage space for WIP was at the North end of the buildlng. One class of wheels – which represents 70% of their build – is processed, palletized and stored 4 times before it is completed. A touch diagram illustrated that a minimum of 13 non-value adding touches occurred for every one of these wheels. During the study, we estimated that a brand new finished wheel already had 2.2 miles on it by the time it left the factory.

Because of the existing process systems, it was decided early on that any infrastructure would need to be installed on mezzanines above the forklift aisles. An underlying design requirement was that the customer could revert to manual operations if the automation were to fail for any reason. This complicated the install, resulting in a rather creative conveyor layout to connect to all of the system elements.

This is a foundary with a large finishing machine shop and paint system. It is hot and dirty wherever you go, and there is usually a fairly thick oil film in the air. This not only created an unpleasant environment for installation, it presented many new challenges for automation where dirt and oil film interfered with operation of equipment and sensors. Oil film on the AS/RS rails, for example, has to be cleaned on a regular basis to prevent the cranes from sliding past target positions.

Finally, the most critical aspect of the solution was the part the vision systems played in the overall concept. Without vision, this project couldn't have been built. There is no machine readable identification on the wheels, so vision had to be used to discern what was being handled.

While vision for this purpose is used at other BMH sites like Topy and CLA, it is used for a much smaller set of 5-8 different wheels – all in an "as cast" format. Here, we had to be able to identify 30 to 35 different wheels, some of which vary only in height, and each having the ability to exist in one of three process states (as cast, painted not machined, painted and machined).

The flow rates into and through the system were segregated illustrate the need for dynamic scheduling of the conveyor and buffer resources. Inbound flow of wheels comes from 5 sources. There are two heat treat systems, a return flow from leak test of wheels that have been machined and are ready to be finish painted with clear coat, and two flows of wheels from the paint systems for wheels that are partially painted before being machined.

A key element of the system is the Daifuku storage buffer. These ultra-high speed miniloads are the first of this generation of product to be imported to the United States, and they are installed within an all Daifuku aisle hardware and rack design. Originally, we designed the system with segregated buffers, - one for paint WIP and one for Machining WIP. This proved to be too costly and space consuming. After we were selected as the vendor for this project, we redesigned the system to move all storage into 15 aisles in two buffers south of the building. Now, however, the space was mixed – meaning that paint and machining WIP co-exist in any buffer location. While this lowered the number of storage locations desired by the customer by about 40%, the improved technical utilization of the buffers made it a better and much more responsive design.

The flow path to machining comes from both buffers, with the vision systems sorting out the flows from buffers to machining and paint, and then sorting the flow to machining to the individual machining cells. The inbound flow of wheels is significant. Special application techniques were deployed to accomplish this level of performance with the number of cranes we could physically fit into the space available.

The Machine cells are the bottleneck of the process. They can consume up to 425 wheels per hour. The complexity of the conveyor network from the buffers to each of 10 machining cells coupled with the high movement rate for wheels from the buffers to paint, required that we develop an intricate "demand pull" system for releasing wheels out to this department. To execute this logic, we had to extend control into their existing system, so that we could sense the actual demand.

The highest rate of flow in the system is to the paint systems. A total of 1600 wheels per hour are pulled from the buffer and routed to the load points for the paint booths. Paint can process wheels faster than any other function in the finishing process. Therefore, it is scheduled rather dynamically, based on the real time demand for wheels into machining.

This is the primary technical reason buffers were recommended in the study. Because of yield issues, customer order requirements other process issues, the system needed a method of decoupling the machining and paint systems from each other and from casting.

In the pre system process, this was an open loop decoupling. By this, we mean that fork lifts, human labor and floor space were all used as needed and in a totally ad hoc fashion, resulting in nearly 31,000 WIP wheels as we started the study. With the system in place, the schedulers for all processes have a clear, real time visibility into what is needed, and can plan accordingly – thus reducing total WIP.

Overall, this is a very busy system with simulated utilizations of equipment will into the 80% levels under extreme conditions. Creative control algorithms and application techniques balance the system's performance among all of the competing demands, making sure no opportunity to service a request is lost or wasted.

As mentioned, Machining is the bottleneck, but Casting produces wheels a bit like sipping from a firehose. While casting is scheduled by projected customer demand, process considerations often lead them to extend a production run well past requirements when high product quality is being experienced. This has always been done independently because there has been no real time visibility into the WIP capacity of the process.

The buffers decouple, smooth and coordinate the flow from casting to paint and machining – eliminating the need to manually offload, palletize and then reload wheels. In real time, however, the customer scheduler's can now see and predict when casting schedules need to be modified because of capacity constraints. The BMH HMI allowed the customer to identify errors, Estops, jams and lift faults. Warning messages were added over time to help AAP manage the system. The Exacta HMI was very useful to AAP associates. The % full of each crane was listed along with status state of each crane. The Exacta in put screen allows each wheel to have individual parameters for quarantine, QC, final destination. The AOR is helpful to identify the status of a " wheel call" and cell activity. The inspection screen allows for QC inspections based on wheel ID and time stored. The Exacta HMI was very useful to AAP associates.

The project was initiated with a sales call orchestrated by Rick in Fall of 2006. Steve had worked on this project when he was with Daifuku, but could never get the customer to consider the level of automation that would be used. A new customer team member – the controller of this operation – saw the opportunity and contracted with BMH to consider the design.

As stated earlier, the hardest part of this project was getting the customer to realize that a stable, well designed and structured process would be the key to making automation work and pay for itself.

By Summer of 07, we had a plan and some momentum. We were selected as the integrator in the Fall of 07, and following some major redesign work to shift the layout and break the project into two logical phases, began installation and completed testing in May of 08. Full production had been realized by the time our onsite staff left the project in November 08.

This seems like a long project – and it was. Two years from concept development to beneficial use – but this project has accomplished what no other wheel producer has accomplished in terms of rate, diversity of product produced, and responsiveness to demand changes – all coupled with a near total absence of human material handling or fork lift vehicles.

Installation began on December the 26th. Other than the week over Christmas shutdown the entire system was installed during productions hours. Communication and cooperation was important to keep interference to a minimum. The conveyor systems weaved in and around existing pipes, equipment and building columns.

Overall the manpower reductions for Phase 1 were achieved. You can safely walk the main aisle without the sound of screaming lift trucks flying down aisle.

An internal success, we successfully demonstrated the ability to direc tly control the AS/RS cranes, a function Daifuku has always demanded they do themselves – at a significantly higher cost.

The dedication of the team impressed the customer, who is now asking us to get the next phase of this project underway by 2nd qtr this year. This project has a lot of visibility within the Hitachi Metals group, and has clearly established BMH as an expert and innovator in the field. Most importantly, this project's success has improved our customer's ability to succeed in a very competitive industry.