Internet of Thing (IoT) in manufacturing industry

backendcode

Member
Join Date
Aug 2017
Location
brampton
Posts
249
Hello Everyone, I would like to talk about Internet of things in Manufacturing industry and would love to know your opinion because I have plan to start my company to implement IoT which will help production lines or integrator to installed a system that can collect production data and useful information for decision makers.

First of all, I have been in manufacturing industry since 2012 and I am pretty sure there are people on this forum who might have experience more than my age and I would love to hear from them as well

I worked as a CNC programmer and then eventually moved into control engineering where I am getting an opportunity to program PLC and Robots which is great but after working for a while in manufacturing industry, one thing I noticed is "LACK OF INFORMATION FROM THE FLOOR" such a downtime, cycle time, run time, production data, most importantly the reasons behind downtime etc etc

Currently, I work with seniors engineers/programmer and with due respect, they hate that data or floor information and their goal is to build a line, finish programming and keep the line running and if stops, automation technician or maintenance will look into that and fix it. They don't want to write a code to collect the data to see the efficiency of line they just build.

And I believe, displaying production (How many parts made) on HMI is not enough.

On the other hand, Management in most of the industries are business people and they just know the "NUMBERS" and that's all. They are not very technically sound and doesn't have much information about what kind of data we can get from Automation line for future improvement (Not talking about engineering or maintenance managers)

Honestly, I love data and information collection from production lines and i believe building a production lines and make them running is only 50% and rest is to improve the line and reduce the downtime, and I think we can have accurate data by collecting that data from automation line, stored on server and give access to decision makers who can look into that data and do something

for example, Machine 1 on production line has same downtime as Machine 2 but Machine 1 made double parts than Machine 2. WHYY? Is Machine 2 cycle is longer than Machine 1 or Robot priority to feed a parts to both machines are not good enough or effective that's why both Machine can't be use in most effective way.


My Question is what do you guys think of collecting data from production lines and stored them on the server and use Big Data tools to make an effective reports from them and benefits company.

I personally believe, there is so much need of this data collection on big level, not just having a production numbers with hand written downtime which was guessed by Operators just to give supervisor because he requested after every shift.

Thank you so much for reading long post and what do you guys think about it?

I am looking forward to hearing from you and I am sorry if I hurt someone with my post.

Thank you,
Backendcode
 
You are correct... vertical data is missing in industry.

However, none of the new buzzwords like Industrie 4.0, IIOT, etc... address the fact that in a lot of places, the data is there but managers are either ignorant or above going to one of the production screens to see it.

Tell me what do you offer that a process historian and a set of properly designed reports can't offer?

I've asked this time and time again to people pushing for all the latest buzzwords and none can give me an answer.

In places I worked either we managed to get a ton of information on downtime by pulling historical alarms into Excel and tackling numbers first or by going back on a batch asking questions why nothing happened in a reactor during X period of time. In one of the cases, the gain was simply mind boggling and all with a simple look at a trend.
 
I’m surprised you haven’t seen more of this.

BI (business intelligence) and Business Analytics is a huge deal for any enterprise deployment. Maybe the jobs and areas you have worked aren’t at that level yet. The cost can be huge, but cloud solutions have substantially lowered the cost compared to on site deployment of BI systems. Companies just have a bad taste of piping out their data to this magical area that most IT people really don’t understand.

Another hurdle is how do you send out data from the PLC and get it to a server, wether local or cloud based, you still need to figure this out. Your SCADA system is a good choice, OPC server, or even physical devices can all help. Once data is stored and formatted in a database, and you have all necessary data points, then you just need a front end to connect the user to the data.

This should almost be mandatory as I don’t see how a company can be competitive without goals and metrics. Now as I said earlier, this is not something new and there are a lot of companies that offer solutions for this. Lots of logistics companies have proprietary software to handle this, but they all still do the basics, collect, compile, and display data.
 
Forget all the fancy names like IOT, big data, and industry 4.0

And I believe, displaying production (How many parts made) on HMI is not enough.
What would you do? What data would you collect? How would you use it?
I have been involved in a couple of data collection projects. Often we got more data than what was used or useful. They had big data and didn't know what to do with it because it wasn't defined. The big data just filled up disks.

The definition is key and how it will provide a return on investment. Collecting data is easy. Finding useful information from it is more challenging.

Useful ways to use information.
1 allocating resources between plants.
2. comparing plant production.
3. comparing shift production.
4. finding bottlenecks.
Then you answer why. When aren't resources allocated correctly. Why are there difference in production. Why is there bottlenecks.
 
You are correct... vertical data is missing in industry.

However, none of the new buzzwords like Industrie 4.0, IIOT, etc... address the fact that in a lot of places, the data is there but managers are either ignorant or above going to one of the production screens to see it.

Tell me what do you offer that a process historian and a set of properly designed reports can't offer?

I've asked this time and time again to people pushing for all the latest buzzwords and none can give me an answer.

In places, I worked either we managed to get a ton of information on downtime by pulling historical alarms into Excel and tackling numbers first or by going back on a batch asking questions why nothing happened in a reactor during X period of time. In one of the cases, the gain was simply mind boggling and all with a simple look at a trend.

Hey cardosocea,

The most important thing I believe is to capture the important information from the production line and what information will help us in future if we use that information is the proper way. for example,

1. Predictive maintenance, Calculate the time for regular Maintenance to avoid a future breakdown in Machines. I know some machines don't get touch by maintenance if they are running good, how about keeping the record of regular maintenance to increase the life cycle of a machine and send a notification for regular maintenance. This will be a very important piece of information the in the long run


2. Inventory Managment, I work for a pretty big company (Canada's number 1 automotive company) and I travel to different plants to support the production line and I have seen many times they are not running the line because of no raw material or sometimes supervisor look for raw materials in the plant and not sure if they have any containers left of raw material but IoT can keep the track of raw material, Pretty sure they have record of pieces in the paper but how much parts were feed on particular line this month? so which can tell if something left in a plant, maybe keep a record of that too? very valuable information right?

3. Quality Control and record, Well, Great we achieved the protection goal but how many parts failed the quality test, how many of them needed a rework etc etc , record detailed information about it.

4 Some of the data collection will be standard and will need on every production line but some of the data collection will depend upon the line and the outcome we want, and then once we have data, use Big data tool to predict and improve things.

Most of the big companies have their Data scientist who collects the data to predict the customer behavior, so I think capturing the right data and use them in a right way to get something out of it is the key point of this IIoT thing.

But using a right algorithm and tools to predict the future behavior and use that data will be challenging but definitely a game changer in this industry.

Please let me know what do you thinnk?

Thank you,
 
I’m surprised you haven’t seen more of this.

BI (business intelligence) and Business Analytics is a huge deal for any enterprise deployment. Maybe the jobs and areas you have worked aren’t at that level yet. The cost can be huge, but cloud solutions have substantially lowered the cost compared to on site deployment of BI systems. Companies just have a bad taste of piping out their data to this magical area that most IT people really don’t understand.

Another hurdle is how do you send out data from the PLC and get it to a server, wether local or cloud based, you still need to figure this out. Your SCADA system is a good choice, OPC server, or even physical devices can all help. Once data is stored and formatted in a database, and you have all necessary data points, then you just need a front end to connect the user to the data.

This should almost be mandatory as I don’t see how a company can be competitive without goals and metrics. Now as I said earlier, this is not something new and there are a lot of companies that offer solutions for this. Lots of logistics companies have proprietary software to handle this, but they all still do the basics, collect, compile, and display data.

Hey Maxkling,

You mentioned very valid points and I agree with you that business intelligence is a very big deal but I have a general thought in the manufacturing industry who use automation for their manufacturing process.

First of all, let's not forget if any manufacturer using automation Robots, PLC etc, they will be either medium size or big companies where they definitely need a data and information collection automatically and in a very detailed way to improve.

I know someone and their industry start using redundant PLC in their control panels and the only purpose to that PLC is to collect the data and stored information, Now there are certain drivers and server who can talk with field devices and hardware and collect the information as you mentioned OPC is one of the communication protocol to talk with Industrial hardware, now once you are able to talk with devices and read the information, you can move that data anywhere you want it, AWS Cloud, Your own server etc and you can create GUI application to represent that data in most effective way and ofcourse do your data analysis to predict and improve prodction lines as well.

you are correct, This should almost be mandatory and I agree there are many industries who might offer such service but just wondering why they are not implemented that much, and if it is then at very limited level. Do you think working with integrators and offer this service as an add-on to production line would be a good idea to get accepted by the customer and they agree to get detailed information about their production lines?

Thank you for your input :)
 
What would you do? What data would you collect? How would you use it?
I have been involved in a couple of data collection projects. Often we got more data than what was used or useful. They had big data and didn't know what to do with it because it wasn't defined. The big data just filled up disks.

The definition is key and how it will provide a return on investment. Collecting data is easy. Finding useful information from it is more challenging.

Useful ways to use information.
1 allocating resources between plants.
2. comparing plant production.
3. comparing shift production.
4. finding bottlenecks.
Then you answer why. When aren't resources allocated correctly. Why are there difference in production. Why is there bottlenecks.

Peter Nachtwey, You are absolutely correct! Even though I have been involved in some sort of data collection projects and once you have data, how do you use that data is very important.

Data can be defined as a KB/MB/GB or it could be defined as a useful information which can solve and predict mistery.

That's what Data scientist involves, For example, I have a friend who is a data scientist for retail business and they collect so much data to predict the customer behavior.

For example, on a certain day, Store sells 20 shoes and 15 socks! and these are just a numbers/data right? but we can predict so much out of it

How many customers bought shoes and Socks together? Do we need to put the socks beside the shoes? because customer behavior is customer always wanted to wear new socks with new bought shoes etc etc

But if you look these numbers on excel sheet, they are just numbers like 20 shoes and 15 socks sold on that day and this much profit was made.

So I think collecting the data (important information) and analyzed the data could be very challenging but I also think Operators behavior is also very important to know who is working on lines, I might sound crazy or be micromanaging but I think it is important to know operators behavior as well like how often button is pressed, when was the first part loaded by operator at the beginning of shift, for example shift starts at 7 am but operator load first part at 7:20 am.

long story short, collection and prediction of data depends on the goal and production line and what we want to get out of it


Thank you so much for reading and replying!

Thank you
 
I think your uphill battle will be working with the local IT group. Some smaller companies won’t be too bad, but there isn’t any real money there either.

I have wrote extra code in machines to calculate and trend production metrics. Guess how often they are looked at? Never in a production meeting is it brought up “the data on the hmi concerned me about last shift”. They have paper for logging issues that then get filed in a filing cabinet never to be seen again.

Perhaps a better service you could provide is taking information they already have and making it better or easier to lookup.

Maintenance reminders on machines are sort of deprecated in my opinion and annoying. Telling me a sensor reading has been fluctuating lately is useful. A motor is drawing more amps than usual is useful. It’s time to grease the bearings and change the air filter, Yeah yeah Yeah I know. I can’t PM a machine that’s in operation. Most machines I work with are run-to-failure though.
 
Go for it. Now is the time to ride the buzzword wave to the bank. Not tomorrow. There is nothing new, arguably not much better overall, but that’s the way it’s always been.
 
Hey cardosocea,

The most important thing I believe is to capture the important information from the production line and what information will help us in future if we use that information is the proper way. for example,

1. Predictive maintenance, Calculate the time for regular Maintenance to avoid a future breakdown in Machines. I know some machines don't get touch by maintenance if they are running good, how about keeping the record of regular maintenance to increase the life cycle of a machine and send a notification for regular maintenance. This will be a very important piece of information the in the long run


2. Inventory Managment, I work for a pretty big company (Canada's number 1 automotive company) and I travel to different plants to support the production line and I have seen many times they are not running the line because of no raw material or sometimes supervisor look for raw materials in the plant and not sure if they have any containers left of raw material but IoT can keep the track of raw material, Pretty sure they have record of pieces in the paper but how much parts were feed on particular line this month? so which can tell if something left in a plant, maybe keep a record of that too? very valuable information right?

3. Quality Control and record, Well, Great we achieved the protection goal but how many parts failed the quality test, how many of them needed a rework etc etc , record detailed information about it.

4 Some of the data collection will be standard and will need on every production line but some of the data collection will depend upon the line and the outcome we want, and then once we have data, use Big data tool to predict and improve things.

Most of the big companies have their Data scientist who collects the data to predict the customer behavior, so I think capturing the right data and use them in a right way to get something out of it is the key point of this IIoT thing.

But using a right algorithm and tools to predict the future behavior and use that data will be challenging but definitely a game changer in this industry.

Please let me know what do you thinnk?

Thank you,

Hi, I think perhaps I didn't pass my message across.

All the points you raise are valid and I agree with you. My pet peeve is that companies are pushing this wave of IOT for something that has always been possible if you have competent people and a data historian.

Like you say, if a company has a data analyst and good data recording/reporting, why would the company need to invest in a tool made for all (not specifically for the business), connected to the internet (which has its own downsides) and a lot more expensive than something that tends to exist already.

Also, there is never too much data, there is either sensors in bad places or not enough questions being asked. And the reason for this is that each department in a plant has different data set requirements and objectives. As an example, Production usually doesn't care about alarms... process safety will be all over the number and type of alarms and maintenance will want to know when maintenance related alarms come up.

Likewise, maintenance doesn't really look too much into analog readings from most process instruments, but if they have motor currents, etc... they'll be all over it, but no other department will care about it.

Edit:

However, if this is what it takes for decisions to be based off of data, so be it... but I don't think many people will get benefits from this as the mentality must change in companies first. Not the tools.
 
You have to show people what's available not tell them. I led the project to implement IoT (before it was called that!) on one of my company's larger products. At the beginning I kept telling the Tech Support people all that could be done but they didn't care. They had no idea what they would do with all that data. All they wanted was the same pieces of data they were already getting via a modem connection.

My team had to present the data to them before they realized how useful it was. And once they had it, they couldn't imagine how they lived without it before!

I knew a manufacturing engineer who needed certain metrics from his production line so he built a simple VB app to gather it and put it up on a big screen by his desk. Same thing happened: once management saw what was possible, they loved it and wanted more. Telling them that the data can be collected is meaningless until people see how useful it really is.

You don't have a technology problem, you have a sales problem. Show them, don't tell them. You have to build a pilot and once people see what's possible, then they understand.
 
You have to show people what's available not tell them. I led the project to implement IoT (before it was called that!) on one of my company's larger products. At the beginning I kept telling the Tech Support people all that could be done but they didn't care. They had no idea what they would do with all that data. All they wanted was the same pieces of data they were already getting via a modem connection.

My team had to present the data to them before they realized how useful it was. And once they had it, they couldn't imagine how they lived without it before!

I knew a manufacturing engineer who needed certain metrics from his production line so he built a simple VB app to gather it and put it up on a big screen by his desk. Same thing happened: once management saw what was possible, they loved it and wanted more. Telling them that the data can be collected is meaningless until people see how useful it really is.

You don't have a technology problem, you have a sales problem. Show them, don't tell them. You have to build a pilot and once people see what's possible, then they understand.

I agree with this 100%, most of the ideas I have implemented were never mentioned during the initial set-up meetings or brainstorming phases. I would just create something and showcase it and then people would like it and encourage the same ideas on future machines. It's easy for us to just give our bosses what they ask, but at the end of the day, a good engineer can come up with new ideas and implement them without anyone asking for it to begin with, while at the same time maintaining other obligations on the project.

This was a good reminder for me as well, great post OP.
 
...I have plan to start my company to implement IoT which will help production lines or integrator to installed a system that can collect production data and useful information for decision makers.

Your proposed means of data collection implementation will probably scare the daylights out of the upper management; and that's because the resulting databases, once connected to the 'Internet' (or residing on the web/cloud), will be accessible to other 'parties', not necessarily the ones interested in the well being of your company.

Floor data collection had been implemented for the first time more than thirty years ago and pretty much any serious business has had some sort of Production/Efficiency/Quality (MES) database going on for many years.

PI (Process) databases collect time-series data (each data point is recorded with a time stamp) while Relational databases (SQL type) collect data and deposits it with user defined archive 'structures' (columns and rows) for easy retrieval/reference; any major industrial software developer has at least one product dedicated to one or both of the mainstream database collection methods.

The most important difference between the currently accepted data collection implementations and your proposal is the fact that the existing databases are located and function within a strictly restricted, proprietary and heavily guarded local system.

I personally do not foresee any IoT proprietary and highly sensitive data collection anytime soon. It doesn't matter if a database resides on the Microsoft Azure cloud servers or on Amazon's AWS ones; if the data is being collected over Internet/web it could be easily hacked by any 'interested' parties.

Pitch your proposal accentuating the benefits of the data collection implementation while firmly stating that the databases will be located within secure and highly-available company assets (servers); do not ever mention any concepts such as 'Internet', 'Cloud" or 'Web'.
 
We have a complete system here.

our comms is via fiber and is fault tolerant. from the switches we do cat 5. if fiber 1 from switch 2 to3 goes bad, there is a fiber from another switch to communicate to it.

we use SQl to interface the plcs to the report system.
SQL communicates with the plc's and puts date in the respective table.
a report generator puts the information in report form at a specified time interval.

from the IT department offices (we are the plc programmers), we can remote to almost every plc in the plant. all pc's, so we can see what the operators are having trouble with.

A word of caution!
have 2 networks at all times. #1 is the corporate / management side that sees the data from the reports. #2 is the plc side and management cant get to it.

the reason is a firewall and to keep management from getting into the plc programs and changing the program / timers, specification limits without knowing what the result is or who is working on the machine.

we can remote in from anywhere to see what is going on, BUT, when any changes are made, we have a strict procedure to follow to keep everyone clear when the changes are inserted into the program.

james
 
Your proposed means of data collection implementation will probably scare the daylights out of the upper management; and that's because the resulting databases, once connected to the 'Internet' (or residing on the web/cloud), will be accessible to other 'parties', not necessarily the ones interested in the well being of your company.

Additionally, some managers that need to be massive stakeholders of the project won't want his managers digging through the data and questioning them.

A word of caution!
have 2 networks at all times. #1 is the corporate / management side that sees the data from the reports. #2 is the plc side and management cant get to it.

I would say three networks... corporate, SCADA and PLC's. But it will depend on layout.
 

Similar Topics

This is more of a multipart question involving home networking and I have limited experience in computer networking. The first objective is to...
Replies
0
Views
295
Engineers: So, we are supposed to do some mods to the controls at a pumps station with a SLC 5/03 and it needs to communicate with another new...
Replies
13
Views
2,411
I have a pc running windows 10 pro that is on a network that does not have internet access. I need to put RSLogix 5000 on this computer for a...
Replies
14
Views
4,659
Hey all, This question is identical to questions that have been asked by many many others before me. I am not happy with the answers i found over...
Replies
18
Views
6,807
i have siemens MP377 HMI which is connect to siemens S700 plc. Now i want to connect HMI with Internet so my supplier will modify my programe.so...
Replies
1
Views
1,065
Back
Top Bottom