Home > Community> Forums
Topic Title: RFID and Pick-to-Light in Lean Manufacturing
Topic Summary: How to construct an RFID system that improve the performance of the production line
Created On: 01/16/2011 06:29 PM
Linear : Threading
Send to a Friend Send to a Friend
Search Topic Search Topic
Topic Tools Topic Tools
View similar topics View similar topics
View topic in raw text format. Print this topic.
01/19/2011 08:58 AM
Print this message

Author Icon
WandiLin
Wandi Lin



Dear all:

Let me say hello to everyone first :-)

I am a new bird here and this is my first post at Lean Forum. So please feel free to note me if I say anything wrong in my message, thanks!

I am looking for advices about how to build a RFID system to improve the performance of a production line and achieve lean manufacturing. Also I am interested in your experiences of constructing RFID and pick-to-light system in your manufacturing systems.

I guess while doing LEAN, many people here have developed RFID systems or Pick-to-Light systems to improve the production processes. Currently, I am in a student team and we are going to develop an RFID based production management system to improve the performance of a shoe production line. In our plan, the system will mainly consist of 4 parts--RFID information collection system, software control panel, a database, and a pick-to-light system which can be triggered by the software.

The desirable functions of this system mainly include:

1. Read the information from the tags attached to each pair of shoes when it passes every production section. Track the location of each pair of shoes.

2. With the information from the readers, calculate and record the performance of the production line. (Cycle time of every order, WIPs, operation times of the workers, etc) Then the information can be used to improve the system, such as determine settings for Kanban, etc.

3. With the information read from the readers, trigger the pick-to-light system; give signal to the workers by lighting up an LED (tell the worker which component to use, where is the component).

The main problems we are facing include:

1. What kind of RFID readers should we choose, considering both the cost and the performance? (High frequency 12.56MHZ or UHF 900+MHZ)

2. Any experience in constructing a pick to light system? Shall we use single-chip-processor (I've tried Arduino, performance is O.K.) Is there any existing chip/platform for pick-to-light system?
01/21/2011 04:48 PM
Print this message

Author Icon
Robert_Simonis
Robert Simonis



Wandi,

I advise you proceed with caution and use a formal project management process with prototypes and a pilot program before widespread implementation and investment. I continue to watch the literature on RFID but, when we piloted some RFID applications in our receiving department about 5 years ago, we were only able to acheive 99.4% accuracy. Anything less than 99.999% is unacceptable given the tens (or hundreds) of thousands of parts moving by the scanners every day.

Many multiple reads, "shadow" readings, blind spots, missed reads, etc. The bandwidth, hardware, and computing power required to get what we needed exceeded our savings from reducing the time and labor needed to receive and input material with barcodes.

I was on the cutting edge of technology and got cut. RFID is a very promising technology but not ready for my industry, yet.
01/21/2011 04:48 PM
Print this message

Author Icon
Running_Lean
Dan Barch



Wandi,

First, I am familiar with Shoe manufacturing, so understand your context.

Second, you sound like a person with a hammer looking for nails. I understand the use of RFID but it sounds like a thousand dollar answer to a two dollar question. A shoe should spend no more than a few hours in the production system, unless you have a lot of WIP - so the goal is to eliminate the WIP, not track it (tracking is Waste).

OK, Toyota uses RFID on their car line. Cool. Tens of thousands of parts. Product line is kilometers long. Communication is difficult. Shoe lines are (should be) less than 100 meters (been there seen that). Maybe 50 parts, usually half that number. Unless you're still doing the 15 second cycle (waste of motion in every step!) you have under 120 people in the line. Why RFID? Get a visual solution. If you have to have RFID to find a pair of shoes, you have way too many in process.

If you really understood the technology you would not be asking this question. If you don't understand it, how can your operators understand it? If they don't understand it, how will they use it? They will not. Been there, seen that too.

Tell the manager asking you to do this to write to me, I know of a Lean program in your business. If this is your idea, it might be a good investment.

Dan Barch
danbarch@ymail.com
01/21/2011 04:48 PM
Print this message

Author Icon
AerospaceLean
Eric Chan



Wandi, have you read about what Kiva did for Zappos.com? You can google it - I think this is the system you are referring to.

eric@aerospacelean.com
01/24/2011 10:01 AM
Print this message

Author Icon
WandiLin
Wandi Lin



Thanks for your reply Rob.
I fully agree that RFID applications should be implemented carefully step by step.

As you mentioned, "many multiple readers "shadow" readings, blind spots, missed reads, etc" What do you think is the main reason which caused these problems?
If the quality of the RFID reader is the main reason, which coefficient of RFID readers is the key? (Frequency?)

Thank you very much!
01/24/2011 10:02 AM
Print this message

Author Icon
WandiLin
Wandi Lin



Dear Dan:

Thank you so much for your kind reply!

Yes, currently, RFID's main function is identifying/tracking, and tracking the order sounds pretty unnecessary in a shoe production line. However, in my opinion, the function of tracking things can be integrated with some other function, and become useful.

First, by tracking the shoes, RFID can trigger the pick-to-light system to let workers pick out components quickly. (In some sections of our customized production line, workers have to pick our 1 from 20-30 components which have different color, size. However, they are difficult to be identify be eyes). This will hopefully eliminate the searching time and mistakes.

Secondary, RFID readers can automatically record information of the production line, which can be useful for continuous improvement to achieve lean. By RFID network, we can record, for example the operation time of each worker/section, WIPs in each section, and store them to the database. By statistical analysis, they can tell us about the performance of different sections of the current production line and show some potential problems.

For the cost problem of RFID system, the high frequency RFID readers have reasonable price near $200/reader in USA and even lower in China. We also considered UHF reader but guess they are unnecessary and too expensive.

Thank you for suggesting the lean program, it is what we need. We already have another group of engineer doing lean manufacturing projects to improve the production, and I think that a manufacturing management system may help them with their tasks.
01/27/2011 03:46 PM
Print this message

Author Icon
Running_Lean
Dan Barch



Wandi,

Thank You for your patience.

I understand the use of the RFID that you are planning. Selection of different colors / sizes using parts is difficult. RFID has been shown to be useful in kit picking where the appropriate part bin lights up to be selected.

However, that is not a production line, and what is sounds like is that you are 'automating the waste.' Have you considered how to eliminat the selection / chioce at point of use? Those car lines I mentioned are supplied in a FIFO manner, the operator only chooses the next part in line which has been coordinated to match the car in front of him. Failure to match causes an andon pull. Value added operator does value added work, support (material supply) does material coordination. Admittedly I have not seen this work for one pair shoe flow, but for 12 pair batch it is easier.

Not saying that what you are doing will not work - but it will take a lot of maintence on your part. The advice here is to engage your operators minds and do some good interactive gemba / team time before investing in automation and deep statistical analysis in a process that may not be stable yet. (statistics are for polishing a process, not for the rough grinding).

"...group of engineer doing lean manufacturing projects" begs the question of operator training and involvement. No project or technology I have ever seen has lasted long without operator ownership. been there seen that tooooo many times.

Good luck! Keep us posted!
03/01/2011 11:54 AM
Print this message

Author Icon
Jalta
John Altamirano



Wandi,

Check out our assembly performance management system. Reading the posts, It may help solve some of the challenges you are dealing with.

www.perfectpart.com

Good luck!

John
04/27/2011 10:53 PM
Print this message

Author Icon
256009
Sangeeth Priyanath



I'm really happy to join this forum as a newbie because there are so many good discussions and I can get good knowledge from those.
http://leanmanufacturingshop.com/
02/29/2012 04:01 PM
Print this message

Author Icon
272161
Joe Pelej



Hello,

I'm with a company called Lightning Pick, we have built Pick-to-Light systems for various manufacturers, some of whom elected to use an RFID tag to launch the Pick-to-Light sequence. Others have initially pursued a more gradual implementation of light-directed technologies, usually doing a Proof of Concept system on one kitting or assembly line, sometimes even just trying it on one small workstation application - say a spare parts kit potentially comprised of configurations of 100 parts or less. We of course do sell a pre-packaged product called Build2Light where a standalone software & database of your BOMs drive the proceess. Usually the work is launched by an inexpensive tethered RF scanner, and a nearby sign with bar codes for each potential build. The operator scans the label and the sequential process starts.

Sales pitch aside, implementing Pick-to-Light or Build-to-Light is a relatively low cost and impact way for production operations to get their toes wet in semi-automated processes. Here's how the Rochester Institute of Technology used both traditional Pick-to-Light (integrated to a host MES or broadcast system) and Build-to-Light (standalone) on 2 workstations in their Toyota Production Systems Laboratory to explore the application of light-directed error proofing in both cellular manufacturing and dedicated parts warehouse style material handling.

http://www.lightningpick.com/InTheNews/ToyotaLab.html
03/02/2012 01:53 PM
Print this message

Author Icon
BradShute
Brad Shute



Hi,
Along similar lines, my current project is investigating the use of barcoding within a sheetmetal jobbing shop (cut and fold only). Initial focus is to integrate with our M1 ERP system to clock and out of various work stations as jobs travel through the system. Any ideas or advice would be much appreciated. :-)
03/09/2012 01:46 PM
Print this message

Author Icon
Robert_Simonis
Robert Simonis



Brad,

Why do you want to clock pieces in and out of erach process or step?

It does not change the product so is inherently non-value added. It also is prone to input error and will lead to tremendous amounts of rework and data correction. Eventualy people will realize it does help them and will abonden even the attempt to do it.

I have seen companies spend, literally, millions of dollars trying to track inventory and machine usage, etc using such command-and-control systems so that their managers and accountants could feel that they had "control". They needed to have computers to tell them what to do because the variation at each step and process was too hard to manage. They were able to then accurately guess how much money they were losing.

The alternative is to have stable amounts of inventory at each point and a system that ensures stability. Money is made by focusing on the flow of material through the process.

Use MRP/ERP to help manage the complexity outside of your plant. The plant needs to be run visually and with stability. Use pull, not the push of MRP.
Note: These forums are moderated by the Lean Enterprise Institute. All posts are reviewed within 24-48 hours prior to appearing on the site. Views expressed in these forums do not necessarily represent the views of the Lean Enterprise Institute.