Case Study: Sun Microsystems | RFID Field Guide: Deploying Radio Frequency Identification Systems

A Case Study on RFID Usage in an Inventory Tracking Pilot

Courtesy: Sun Microsystems

The Client

With more than $11 billion in annual revenue, Sun Microsystems is a leading provider of industrial strength computing systems consisting of server, software, storage, and services. The company prides itself on providing innovative solutions to its customers that reduce computing complexity and lower the overall cost of ownership of computing infrastructure for its customers.

The Challenge

Because Sun designs and builds the majority of its servers, efficient supply chain and manufacturing systems are critical in keeping its operational costs down. Higher operational costs can either make the product less competitive or hurt margins. The company has maintained an on-going focus on streamlining its operations and pursued an aggressive program for process improvement, modeled after the Six Sigma methodology.

Some time back, Sun's manufacturing plant in California was looking at various ways to further streamline its operations. This plant was responsible for making some of Sun's mid-range servers and storage systems. Although the manufacturing and sourcing were highly streamlined, it seemed to the plant supervisors that the processes for overall inventory tracking could be improved. RFID technology seemed to be an enabler for this. At the same time, Sun was quite deeply involved in the emerging area of RFID technology through its leadership role in developing standards, as well as middleware software, for RFID. It seemed that a hands-on RFID project to improve internal operations would help Sun stress test the newer versions of its software and create an internal competency of RFID solution architecture and deployment. With these benefits in mind, the company decided to deploy an RFID pilot in its manufacturing operations.

Sun faced several challenges in the selection and design of the RFID pilot. Manufacturing of a mid-range server, the type being manufactured at this plant, requires many distinct steps such as component assembly, testing of sub-assemblies, software installation and customization, and testing of the final product. A pilot that puts RFID tags on every sub-assembly of a machine would require major changes to the manufacturing process and as a result, was considered out of scope. There was another issue as well. Because the pilot would have a fixed start and stop date, followed by a review and possible process improvement phases, if Sun started tagging its production machines during the pilot phase, some customers would have machines with RFID tags and others would not (after the pilot ended). Sun needed to find a pilot in which the tagged objects would remain inside its factories (also known as a closed loop process) and still provide meaningful enough results to make extrapolations about operational savings of a fully deployed RFID system. It found the perfect pilot in its Rotational Capital Process.

What Is Rotational Capital Process?

Testing a mid-range server (multiprocessor) requires a variety of test components and equipment, including other servers, server chassis, and various I/O cards. The testing could occur at multiple stations during the assembly, and may require more equipment than typically available in the test harnesses at that station. For example, to test a CPU board for a multiprocessor server, a server chassis with a certain configuration of disk drives, memory, power supplies, and CPU boards may be needed. If the right configuration were not available, an operator would borrow the missing items from a pool of capital equipment (known as rotational capital). When the testing was completed, the borrowed capital equipment would be returned to its original location. In this manner, the capital equipment can rotate in and out of the pool.

Though an ERP system-based tracking process was in place for the rotational capital, it was not working well. Sometimes, the operators forgot to return the equipment or made a mistake in entering the data into the ERP system. As a result, data synchronizationreconciliation of data between what was physically available and what was in the ERP systemwas an on-going challenge. Several people were working full-time to resolve the arising discrepancies by going around the plant looking for missing equipment and bringing them back to their proper location. Because these were high-value items, the labor cost was justified in terms of not having to keep excessive test inventory.

Scope of the Project

The scope of the project was limited to showing the viability of using RFID technology in an environment with high metal content, and gaining positive ROI (return on investment). It was broken into several steps:

1.	Define use cases based on existing process discovery
2.	Define the architecture and source components
3.	Build the pilot, including any custom software and physical setup
4.	Deploy the pilot for three months
5.	Evaluate the results

A virtual team, consisting of members from the manufacturing operations, Sun Professional Service architects, and engineers from the RFID software team was put together to drive the project. Appropriate management approvals were obtained and budget was secured to conduct the pilot. Because the goal was to create a system that can be rolled out across Sun's supply chain, a standards-based deployment was required. The team decided to use EPCglobal's UHF specifications to ensure such compliance. The software components were already compliant with Java and Jini standards.

Hardware and Software Products Used

Tags. UHF tags made by Alien Technology (I-tag).
Readers. Stationary readers and linear antenna from Alien Technology.
RFID Middleware. Sun EPC Event Manager.
EPC Information Server. Sun EPC Information Server.
Tracking Application. A custom built tracking application was used to keep a record of location histories of each item and a log of all transactions.
Reporting. Brio reporting tool.
Database. Oracle.
Application Server. Sun JES Application Server.

The Solution: How it Works

The pilot system consists of four components:

An RFID tag per unit of test equipment, containing its EPC, which uniquely identifies it.
Readers and antennae, which receive signals from the tags and map them to specific entries in the equipment database.
RFID middleware, which receives information from the reader, filters it, and delivers to the appropriate software program.
Business processing software, which collects and analyzes the data.

For more information about how these different components interact, see Chapter 3, "Components of RFID Systems."

First, all the test equipment was tagged with Alien Technology's UHF I-tags. Because quite a few items had heavy metal content, the tags were affixed on top of a plastic, which was attached to the metal. Linear antennae were chosen due to their enhanced read performance, requiring that tags be vertically aligned to the antennae. The tags were carefully placed to ensure this orientation. Each tagged object was associated with an EPC and entered into the corresponding Oracle database repository.

The pilot layout was created such that the operators taking equipment from the central location to their station or vice-versa passed through a reader gate. The tagged object(s) were identified at this location. The operators were trained to verify the data read by the reader, manually override any inaccuracies, if needed, and complete the transaction. The transaction triggered several events. For example, the location of this object would change in the central repository and the transaction logs would be updated. When the testing was done, the whole process occurred in reverse, where-by the material going through the reader gate was now added to the inventory.

Due to the scope, budget, and duration of the project, the list of tagged equipment in the database was not synchronized with the master list of equipment in the database. It should be noted that such integration would surely be covered in the full deployment.

Results and Lessons Learned

The pilot showed that it was possible to track and trace items with heavy metal content using standardized tags and RFID middleware. Here are some of the findings:

The accuracy of the system was quite high, approximately 99.5%.
Operator training turned out to be an issue. The tags were read properly as long as the operators walked through the RFID gate, which did not always happen.
Linear antennae provided much better response than circular antennae. Tag orientation vis-á-vis the reader was proved critical in such situations.
Although the pilot project touched upon a very small part of the overall supply chain and manufacturing processes, the ROI analysis showed that if fully implemented, the project would have a positive ROI over a three-year horizon, with break-even point in approximately 2 years.
Handheld readers were desirable as they could help track inventory in situations where operators took the test components without passing them through a stationary RFID gate.