A simple sensor for keeping belts moving

[NitrousWolf]

I just felt a little proud about my smeltery for a moment sitting back admiring it so I thought I'd start breaking it down into bitesize chunks and explain them (showing it off basically ).

Below is one chunk I call The Blockage Sensor.
It samples the flow of product on a belt and if it detects a blockage outputs a circuit signal via a smart chest.
If the blockage is removed then the circuit signal is cleared.

This is useful in my smeltery because if both lanes of product on the belt are blocked then a series of smart inserters dump the product into storage banks thus unblocking the belt and allowing the smelters to work on all available ore non-stop.

The Blockage Sensor

SensorA.jpg (585.92 KiB) Viewed 10156 times

[DerivePi]

Have you considered an ->>>>underground belt | slow inserter | smart chest | fast inserter | Underground belt return >>>>>? If the smart chest has more than say 100 items, you know the belt is jammed. Also has the advantage of providing a small buffer.

[XKnight]

Have you considered an ->>>>underground belt | slow inserter | smart chest | fast inserter | Underground belt return >>>>>? If the smart chest has more than say 100 items, you know the belt is jammed. Also has the advantage of providing a small buffer.

The first question you need to answer is "what means belt is jammed?".
If your factory consumes 1 item per 60 minutes is it jammed or not?
In my opinion until you have this definition it is senseless to build any sensor.

Also, this system perfectly handle your task without any inserters:

1.png (177.55 KiB) Viewed 9969 times

[Kelderek]

If your goal is to have a free flowing belt and any excess items go to storage, why not just have a bunch of inserters extract everything from the belt into storage at the very end of your line?

I like your design, I'm not sure I get why it is necessary. Is there a particular reason why you want to dump excess to storage in advance of your main factory instead of at the tail end?

[NitrousWolf]

Thanks for the responses guys. I don't often talk about what I'm doing in factorio so it's nice to get a few different points of view on stuff

Have you considered an ->>>>underground belt | slow inserter | smart chest | fast inserter | Underground belt return >>>>>? If the smart chest has more than say 100 items, you know the belt is jammed. Also has the advantage of providing a small buffer.

I like it. Its a very compact design. Functionally the same as what I'm trying to achieve here so thanks for that I will definitely give it a go.

Have you considered an ->>>>underground belt | slow inserter | smart chest | fast inserter | Underground belt return >>>>>? If the smart chest has more than say 100 items, you know the belt is jammed. Also has the advantage of providing a small buffer.

The first question you need to answer is "what means belt is jammed?".
If your factory consumes 1 item per 60 minutes is it jammed or not?
In my opinion until you have this definition it is senseless to build any sensor.

Also, this system perfectly handle your task without any inserters:

I see what you did there "senseless to build any sensor"
My factory output fluctuates wildly as my activity changes, for instance sometimes I am burning through the research tree or churning out laser turrets, solar panels and accumulators and other times I am almost idling because I'm busy with other less demanding things like railways.
It is during these idle output periods I want to prevent the furnaces to be idling because that means a lot of wasted productivity time.

To keep the furnaces running I have to store excess iron plate
But during peak load I want to be releasing iron plate

Although your image will siphon off any excess plate for storage I'm not sure that it will give a way of releasing it back into the output stream during peak load. Also it looks like it constantly siphons off 1/32th of production for storage am I right? (I am assuming storage is the left hand belt and the output is the right hand belt)

If your goal is to have a free flowing belt and any excess items go to storage, why not just have a bunch of inserters extract everything from the belt into storage at the very end of your line?

I like your design, I'm not sure I get why it is necessary. Is there a particular reason why you want to dump excess to storage in advance of your main factory instead of at the tail end?

In short - space constraints.
The output of my furnaces is a branching mess of belts intermingled with a branching mess of railway track surrounded by lakes (I like high terrain segmentation and big water maps). I could make a belt run to a storage bank placed nearby but I felt it was easier to store where there was space, which is where its produced.
My goal was to keep the furnaces running full-time. I didn't necessarily want to have a free flowing belt out of the factory since I want the belts leading to the trains to be filled with product, however the belts leading away from the furnaces had to be free flowing.

[DerivePi]

To keep the furnaces running I have to store excess iron plateBut during peak load I want to be releasing iron plate

How about a mass storage solution? All of the ore comes off of the train and goes straight to the furnaces. The output of the furnaces, plate, has a higher stacking value so this is what you store. This is my idea for compact mass storage using branchlines from continuous belts to feed provider chests for bots to place into storage chests. On the other side are requester chests that request delivery of the stored items to place back onto the continuous belt. The only time you build up storage is when the belt is clogged and there is no need for a sensor. Here is a schematic of the storage area (note - I am not showing the storage chests which just need to be placed within roboport range) -



Here is a schematic of the overall area


and here is an ingame image - I have ore being stored on the top (no need to do this) and plates being stored on the bottom.


These were taken from this thread - viewtopic.php?f=8&t=8238&hilit=MASS+STORAGE&start=10

The places where I've needed sensors are where multiple items share the same belt. In that case, you need a way to sense when you have too much of one item and prevent more of it from being placed onto the belt. This happens especially for non robot belt sorters at RR stations.

[NitrousWolf]

DerivePi:
That's certainly one of doing it with no need for a sensor. Your method however will shove half of you plate and ore into storage during peak demand only to take it back out again which is a little wasteful of energy (would you consider using xknights belt/splitter suggestion mentioned in his reply here to increase efficiency during high demand?)

Also speaking of energy use the roboports probably dont help. However I like your use of roboports because you can increase the storage very cheaply and easily by just cramming in storage chests without the need to access them with inserters and belts.
Also I like the fact that you can just throw more robots at it to speed it up! Its definitely a good solution for the late game high flow rate factory since it is so scalable

[mooklepticon]

I really like your #2 step, there. I'd been thinking about an activity sensor and that fixes the problem I'd been thinking of.

My inventory buffer solution is to have 1/4 of my output split into my inventory chests. Those chests then dump back into the line about 10 or so belts later. So, yes, 1/4th of my output spends a little more time on the belt, but if the line backs up, it'll auto dump into the buffer. Why 1/4th? Because 1/2 is too high priority and 1/8th takes too many splitters and space.

[NitrousWolf]

Im glad this discussion helped someone. I won't hesitate to share my next creation

[DerivePi]

Your method however will shove half of you plate and ore into storage during peak demand only to take it back out again which is a little wasteful of energy

Yes, it does add to the energy bill as well as the flurry of robots that have to be active. To prevent this, I'm thinking that an inline sensor would do the trick. Place the inline sensor on one of the belts in the middle of the storage area. As the sensor chest fills up activate more of the into-storage inserters. As the sensor chest empties, deactivate the into-storage chest inserters.

Should work as follows:
full output/demand - inline sensor takes item off of through-belt and immediately replaces it into empty spot. Sensor reads less than 5 items so all of the into-storage inserters are off causing the to-storage belts to back up and all items to flow on the through-belts.

no output/demand - inline sensor takes item off of through-belt but can't put the item back on due to clogged output. Sensor reads more than 20 items (chest is limited to single stack of items) so all of the into-storage inserters are on allowing full flow of items into storage since through-belts are completely backed up.

If you want to control the out-of-storage inserters then we need another sensor before the storage area to gage the input. With high input, out-of-storage inserters are turned off and with low input, they are turned on.

[NitrousWolf]

As the sensor chest fills up activate more of the into-storage inserters. As the sensor chest empties, deactivate the into-storage chest inserters.

Ooh so by having graduated levels of response you could get a more corrected flow to storage rather than flip flopping between storing everything and then releasing everything. Now thats sound engineering.

If you want to control the out-of-storage inserters then we need another sensor before the storage area to gage the input. With high input, out-of-storage inserters are turned off and with low input, they are turned on.

In my factory I actually have the same sensor running both the inserters leading to storage AND the inserters leading away from storage. Since they will both measure the same input. If you just run one set exclusively of the other(reverse the condition of the first group of inserters from the second group of inseters) then you should be fine. However I can't yet think of a valid way to do that with your graduated response idea.

[katyal]

I use a splitter filter(the arrangement of splitters that filters a belt)....100% output unless the belt backs up...as soon as the belt is backed up 100% of output is directed to storage. Side load back on to main line from storage so that nothing moves until main line isn't backed up

[mooklepticon]

I use a splitter filter(the arrangement of splitters that filters a belt)....100% output unless the belt backs up...as soon as the belt is backed up 100% of output is directed to storage. Side load back on to main line from storage so that nothing moves until main line isn't backed up

Can you take some screen shots? I'd love to see this.