July 19 2007 12:03 PM

    During 2006, more RFID projects moved out of the lab and into production than in all previous years combined. Accordingly, the lessons that are filtering back from those trials have less to do with science and more to do with the interaction of RFID with the production environment and the creation of sustainable business cases for wider rollout. With that in mind, we give to you the five most important lessons.


    1. Production Is Not Like the Lab

    In 2004 and 2005, laboratory trials helped companies to understand the limitations of the then-current RFID technology. To some extent, it allowed the industry to begin thinking about how RFID could be integrated into operations. But nothing substitutes for production experience. In laboratory settings, events are planned and usually occur at sub-production rates. Its hard to simulate production, because:

               Production rates are higher and include surges from rush orders that are expensive or impractical to simulate in a test setting.

               Human processes are seldom uniform (think new workers, individual habits).

               Production includes randomness from handling damage, misplaced items, unexpected product movements and equipment failure.


    Aside from these random events, putting RFID into the production line exposes deeper scalability issues. Suddenly, the responsiveness of the entire system comes into play. The same attention that was paid to choosing tags and readers now needs to be paid to the RFID automation and integration layer.


    This RFID automation layer not only needs to have great performance (companies are now demanding and getting service level guarantees on system response), but it needs to be flexible, too. There are some amazing (and sometimes disturbing) observations about what really happens in production after RFID monitoring is added. There will be process exceptions that you did not predict. Oh, yeah every once in a while an item falls off the line, and we just walk it around the other side.


    2. Think Closed Loop

    If you have been waiting for a customer, supplier or logistics partner to put RFID tags on items so that you can take advantage of RFID stop waiting.


    Open loop is a term sometimes used to describe the situation in which one party is responsible for the placement and cost of the RFID tag, while another party expects to benefit from using RFID within its own operations. By contrast, a closed loop application is one in which a company doesnt look outside of its own operations for either tag application or a business case.


    Open loop situations are characterized by calls for standards and supply chain collaboration. While everyone agrees that these things will come to pass, the companies with great business cases right now arent waiting for these to occur. They have a completely internal business justification, which relies on improving their own level of service to customers or the efficiency of their operations.


    Examples of these types of projects include:

    100% shipment accuracy

    Remotely monitored inventory

    Asset tracking across many locations


    These are all applications that are very well matched with RFID: they need to happen automatically without any human intervention frequently part of the problem and they may need to happen in many places at once. In this situation, a human solution is not economical. Rather than have a representative count items at each customer site, a business can install $5,000 worth of equipment that will do it automatically and instantly update the central management systems over the network.


    3. You Dont Always Need a Tag

    If youve been doing mental math to calculate the cost of a tag on every item that you handle (Hmm. $0.10 times 2,000,000 items a year comes to about), then consider some creative alternatives. Companies have successfully built strong business cases in logistics without ever tagging each item. Theres another way.


    These other methods all fall under the general category of inference. These methods dont directly sense the presence of a tag on an item; they infer the presence of the item based on other information. For instance: a tag on a container or conveyance.


    While not as theoretically rigorous as the direct detection of each item, the fact that such a system is less than half the cost of an item-level one more than makes up for the fact that it is only about 90% as dependable. To quote one believer: Before, we had no visibility of where items were until they reached a transaction point way down the line. Now we get a signal. And even though its not perfect, its good enough for us to catch and correct problems that we couldnt catch at all before. Dont let perfect be the enemy of the good.


    4. Tags Need Maintenance, Too

    Its easy to get so wrapped up in launching a big RFID project that certain longer-term maintenance issues escape notice until later in the project. One which deserves your attention in any RFID project is the maintenance of the tags themselves.


    For instance, if you tagged totes or package bins in a distribution center, you should consider these containers to be broken if their tag becomes inoperable. That means that youll need a process of screening them, preferably before loading. Then youll want to divert them to your container hospital for tag replacement. In high-volume lines, you can even find ways to automate this step with high speed tag sensors (not the same as a reader; remember, the tag doesnt work) and tag applicators that are suitable for high rate conveyor systems.


    5. You May Need More Than One Type of RFID

    Finally, as RFID becomes a respected and dependable tool in your logistics arsenal, its entirely possible that youll discover the need for more than one type of RFID tag. Rather than complain about the fact that so many choices exist, embrace the diversity: several of them may deserve a place in your toolkit.


    Heres a real example that uses three different types of incompatible tags in the same supply chain:

    13.56Mhz HF tags, with a range of just inches, which are used to enable item sensing on shelves

    900MHz UHF tags, with a range of twelve feet, which are used to sense containers at dock doors

    Long range battery-assisted tags used to find wheeled carriers in a distribution center


    Its significant to note at least in the above case that none of these could be substituted for another. While some technologies have the potential for greater versatility (near-field UHF, WiFi tags with positional sensing), they are either not yet widely available or not yet sufficiently cost-effective for widespread use.


    More Lessons to Follow

    In a future article, well look at some of the creative solutions that companies have developed in response to the hurdles that they encounter during deployment. Good luck with your future deployments of RFID!


    John Beans is the Vice President of Marketing for Blue Vector Systems, a company which has developed an automation platform that integrates a variety of RFID, barcode, temperature and other industrial sensors with back end enterprise applications. Johns experience spans manufacturing systems at Datasweep and Camstar, networking at SBC Communications (now AT&T) and strategy consulting at Bain & Company. He has degrees from Clemson and Berkeley and lives in Oakland, CA.