Conveyor Control - Window Reservation

[brucechase]

I thought I would chime in here. I do the exact same thing with tires merging onto a main conveyor belt. There is no fancy array or tables to keep track of what is where. The photoeyes on the main conveyor are a little further upstream of the induction point (at least 1 bag length). As a bag moves down the main conveyor, there is a timer actuated by that photoeye that give the 'hole' size. Based on the conveyor speed, I've got it down to the largest tire size (to be the same as a bag). If the hole is big enough, then after a slight delay of the tire passing the photoeye and I induct it onto the main conveyor. Be careful of using retentive timers because of the startup of the conveyor system if on a VFD or softstart (which I suspect is the case). The ramp up time messes with the hole size.

If you want to not starve the last counters, then use a set of extra delay timers in the 1st inductors such that if there is luggage on the last of the counters, then I add a delay equal to 1 or 2 holes sizes depending on how long those photoeyes are block.

Hope this wasn't confusing, I just thought it was getting a little more complicated than it should.

 

[elevmike]

Given the various bag sizes and shapes, & the need to be able to regualate conveyer speed etc. I dont think your going to be able to NOT use a combiniation of encoders and photo eyes.

 

[brucechase]

From what I understand, the window is the same size reguardless of the bag. The conveyor speed should be a constant, just starting and stopping. Elevmike you asked the question and it wasn't directly answered so I am making it as simple as possible. Keep the window size at just over 1 meter and constant belt speed. I have several applications just like that where different size tires are inducted into a constant size window. Just use photoeye and timer and it works great.

 

[appleplc07]

Yes... the windows size should be the same size to keep things simple. I will keep it to be 1 m length. The distance between the check-in counters is only about 3m... how many photocell sensors should i put between the counters ?

 

[elevmike]

My personal opinion:

Theoretically, if the conveyer speed & window size is constant, you can get by without using an encoder. However what happens in real life vs. theory, sometimes isn’t always constant. Placing a single encoder on the main conveyer to create & track windows, will assure a constant quality output regardless of small changes in the conveyer speed that could be caused by a number of factors including mechanical wear, power issues or varying loads. There's just too many things that can go wrong. I wouldnt even consider NOT using an encoder.

 

[brucechase]

You should only need 1 photoeye between the counters. This will be the photoeye before the induction point to see if there is a window (or hole) available.

Is the main conveyor a belt or some sort of roller conveyor? Also, will the motor be started across the line or do you have a soft start/VFD on it?

 

[Doug-P]

how many photocell sensors should i put between the counters ?

It depends. If you could guarantee that EVERY piece of luggage from now to the end of time would enter the takeaway conveyor smartly and when commanded, and within a known time, you wouldn't even need any PEs on the takeaway, except for where it merges with another conveyor. However, not every piece will behave the same. Some will do as they're told and when, while others will a.) take their time getting out; b.) get stuck for any of a thousand reasons; c.) drop pieces of themselves into the works and cause jams; d.) block PE's when PE's really shouldn't be blocked.

Usually, more sensors is a good thing (from a programming perspective the more direct information you can have about what is going on out there, the better). As has been noted before, sensors are cheap compared to the effort involved in doing without them - penny wise and pound foolish, as Ben Franklin would have said.

It just depends.

 

[plchacker]

Nice project.

First, and this has been mentioned a couple times. What about the guy at the end of the line? I suggest a FIFO (First In First Out) to keep track of who gets the next empty space. A FIFO will solve several situations. 1. Station closed. 2.Uneven production between stations. 3. The last station will be granted a space when it needs it. I don't know if this instruction is available for you or not, but it does work, and works good.

A FIFO can be created with latched bits, if the instruction is not available. One routine should determine who gets the next available window.

Using a Photoswitch up stream to determine if the window exists would be better than assuming that the luggage is where it should be (according to the encoder and bit array), even with error checking. This would eliminate trying to keep track of a very long conveyor, with lots of possible hick-ups. You didn't mention possible loading of the belt. How much slower will it be with a full load? Belt slippage as well as baggage slippage could be trouble.

If you are concerned about the price of photocells, check out the Mini-Banner, you can get them in several configurations, and they are inexpensive. Last time I checked, about $90 per. They are very good sensors.

I too, think you should spend some time watching luggage. (Let someone know why you are watching baggage in their airport.) From my time in airports I'd expect a lot of irregular situations, that in truth are quite regular.

Oh yeah, lumber sorters are very impressive. A programmer that works almost exclusivly with sorters, taught me a lot about programming. Even after the Austrians, got through with me.

Eine beer bitti?

 

[appleplc07]

Hi Bruce,

Your recommendation to use 1 photoeye which is at least one bag length away from the induction conveyor. In this case, 1 m away from each of the induction conveyor. And between each counters, I only need 1 photoeye. Did I get this right ?

Each time there is a baggage on the induction conveyor, start the routine to count the hole size of 1m, and if this is available then induct onto the main conveyor.


All,
I am just unclear on the bit about the check-in counters downstream starving of available holes…. FYI, one area has 16 counters and the other has 20 counters which is quite a long one. And according to the specifications, for the 16 check-in counters, the throughput is 45 bags/minute. So, our control must keep in mind this number of 45 bags/minute. Let’s say I designed the system using the photoeyes… etc etc… how can I know that my design can meet this 45 bags/minute throughput ? or at least I know that this design will not have a big adverse effect on the throughput.



Therefore, I think this system requires that we balance out the whole check-in counters induction of the baggage on the main conveyor.



By the way, we do have lots of view on the system to use the photoeyes on the main conveyor. Anybody knows how if encoders are only use instead. Apparently, the normal and standard way of doing this window reservation on the main conveyor (for airport project) is via the encoders only. If you look at the airport main belt closely, I have yet to see any photoeyes on them. It is always done by encoders. Anybody can share this information how the windowing is done via encoders.

 

[elevmike]

By the way, we do have lots of view on the system to use the photoeyes on the main conveyor. Anybody knows how if encoders are only use instead. Apparently, the normal and standard way of doing this window reservation on the main conveyor (for airport project) is via the encoders only. If you look at the airport main belt closely, I have yet to see any photoeyes on them. It is always done by encoders. Anybody can share this information how the windowing is done via encoders.

There's good reason for the encoders in this type application. LOTs of them. Lack of baggage uniformity is just one good reason.
Anyway the encoder would measure and mark the windows as they emerge from the beginning of the conveyer. The best way to conceptualize this would be to take an endless belt and mark it off in measurement units such as cm. For example size your encoder to measure maybe 1 cm on the belt for each pulse. 100 pulses = 1 meter = 1 window. You will also need to know how far it is from the beginning of the conveyer to each check in counter chute (or merge point). You will have a memory stack that is pushed down each timer the encoder counter "marks" off a window in your program. Each merge location would have an assigned distance value based on how many encoder counts it is from the beginning of the conveyer. As the windows are created at the beginning of the conveyer your going to assign them to pick up a bag at a particular merge location based on demand. (1st come, 1st served). So when that window reaches the assigned merge location, the bag waiting at that location will be released on to the main conveyer. In short each window will have two adjacent memory locations in the stack. 1) The count value of the assigned merge location, and 2) the current distance from the beginning. When the two values meet or become equal, tell the bag to release on the conveyer. (during commissioning you might have to adjust the "release value" to avoid collisions.

Presumably the check in counters will have a drop down gate to release the bag. The gate would then act as a slide for the bag to drop on the conveyer via gravity. Somehow at the check in counters, the bags will have to be kept separated so that you drop only one bag at a time on the main conveyer. Or the check in clerk will be limited to only having one bag in the que at a time. One way to handle this is to have a marked conveyer at the check in counter. The clerk will place each bag between the marks. After a bag as been released on the main conveyer, the check in accumulating conveyer will index by one window dropping another bag into the que.

To detect bags in the que at the check in counters, you will need a photo eye. However given the great variety in the way people decorate their luggage you want to use through-beam photo eyes to prevent the possibility of a reflection from the bags that could give you a false negative indicating nothing in the que.

Another opinion: Keep the limit for the bags at one meter, but allow maybe 1.5 meters or more for the windows. Not every bag is going to land on the conveyer in the exact same manner.

 

[elevmike]

One more suggestion:

If I were you I would build a working scale model of this setup using maybe 5 check in counters. It will be MUCH easer to program & debug with everything all layed out right in front of you. Debugging at the last minute at the jobsite is going to put you in a cold panic and prevent you from thinking clearly, whereas debugging in a comfortable enviroment and way ahead of the schedualed compleation date will likely have more desireable results.

 

[plchacker]

One more suggestion:

If I were you I would build a working scale model of this setup using maybe 5 check in counters. It will be MUCH easer to program & debug with everything all layed out right in front of you. Debugging at the last minute at the jobsite is going to put you in a cold panic and prevent you from thinking clearly, whereas debugging in a comfortable enviroment and way ahead of the schedualed compleation date will likely have more desireable results.

Boy is that the truth.

Then you can try both methods. Encoders, and no encoders. Garment bags and dog caddies - rifle cases and rolling luggage with pull-out handles sticking out. Lots of situations.

 

[darrenj]

Wow an airport that cares about the luggage..The last couple of airports i was in the check in clerk pushed a button and the conveyor pushed the bags to the main line..They turned there head to make sure there was room (If you were lucky!!) and the ba was pushed on to the main line..If there was no room the button was kept pressed until the bag either weged itself onto the line or a big enough gap opened up...

The airlines cant be that worried about your bags..Look at the pick up carousel..Bags come out..it dosnt matter if there is already a bag on the carousel underneath....

Sounds like a damn interesting problem..Certainly made me think about all sorts of problems..Specs are one thing real life is another..

If the Boss is trying to cheap out go to him/her with the scenario..bag falls off or gets misplaced on the belt and a jam is created..we get sued...BTW the encoder or 5 more eyes is about 1000$..what should we do?? They might see it in a diffent light..

This being a training or classroom exersise i would go with timers..However this is real life..Spend the extra money!

D

 

[appleplc07]

Hi Elevmike,

I hope I understand you correctly…. Let’s say only the last check-in counter i.e. counter 20 is opened for checking-in and the distance from the beginning of the conveyor to this check-in counter 20 is 20m. Then if there is a baggage on the induction conveyor, it has to wait for a count of 2000 pulses for the window to come in order to inject into the main conveyor ? And the next baggage on the counter 20 must also wait for another 2000 pulses before injecting into the main conveyor. If this is the case, then the time taken to inject the baggage is too long. It should inject on the next available window as soon as possible.

 

[elevmike]

Apple,

You've got a good point. However nothing precudes you from assigning an unassigned conveyer window on the fly if that window hasnt yet reached the merge for the open check in with a bag waiting in it.

My big point I'm trying to make here is to use an encoder on the main conveyer to create and track the movement of the windows. Once that's done, assigning winows more efficently wont be diffucult.