Multi Tank Dipping Process

[rajsiyer]

Hi Guys; I've jiust begun work on a new project: It is a PLC controlled Parkerisation process system. Customer requires that PLC controls the 15 tank dipping sequence process, in an optimal manner so that his throughput is substantially improved. In this system, there are 15 tanks holding different chemicals such as Caustic Soda Lye, Water, Phosphoric acid, Sulphuric acid, etc, etc. A device called a transporter travels overhead in the line of these tanks and is capable of travelling to any position including a loading point and an unloading point. Barrels containing small parts are placed in the loading position. The transporter lifts this barrel and proceeds to the dipping sequence from tank 1 to tank 15. Now the immersion time in these tanks varies frm a few seconds to as much as 5 minutes. The travel times between the loading/unloading station to various tanks are in the order of 15-30 seconds. A stable algorithm/flowchart similar to that used for electroplating is what I am looking for. So I'd be grateful for any help/suggestions from fellows who have seen this kind of action earlier. A million thanks Raj S. Iyer

[TERdON]

I'm actually working with PLCs in the electroplating/galvanization industry. We do what you're wanting to do by hand actually, making an optimized program by hand (a great help by this is doing a time-distance-diagram (tid-väg-diagram) and just visually try to move the transporter around. Short immersion times means just dipping the barrel, long leaving it there for a transporter movement cycle, and really long ones means using two tanks in parallell and serving them only every second transporter cycle). If you'd like to see a bogus diagram as an example, please tell me and I'll upload an example. Then we feed our program (the base PLC program is reconfigurable on the part of the lifting sequence by changing DM values) with our sequence, and basically fiddle around with it until we're satisfied. We know that this isn't optimal, but it works, and doing a calculation in the PLC deciding on the optimal route for the transporter just isn't feasible, IOHO. AFAIK other vendors are doing the same, except the ones who are turning to using a computer for optimizing the sequence. Is the immersion time in each tank always the same, or does different barrels get different treatments? Then the easiest thing to do with identical treatment times is just to use my solution. As the treatment is known before hand, you can optimize it easily. Where it starts to get out of hand is when you get to do several different treatments in the same plant - adding two tanks and some options on the HMI is "an easy thing to do" but unfortunately, in a 100 tank plant with 10 different treatments it gets kind of unobvious which sequence is the optimal, especially when the sequence of barrels isn't repetitive (which it usually isn't). It is actually such a complicated and difficult problem that there is a phd position at my University for trying to apply genetic programming on the problem of optimizing such sequences... Another solution altogether, actually, would be contacting one of the companies that are specialist in this area. At my company we reuse large parts of the system, when a new one is built, so it could save you headaches (but not money).

[rajsiyer]

Hi TerDon Great. Yes Yes Yes.. Please Upload the Time/Distance diagram. It would be a great help. I am working on another approach. The basic principle is..The transporter makes a movement decision each time it puts down a barrel into a tank asking where do I go next ?. In this case, since there are only 15 tanks and only one loading/unloading station each the choice is only 1 out of 15 possible destinations. From any tank to any other possible tank/point we know the approximate time. Next there are tanks which contain acids which can thin down the metal if immersed longer than required. So removing the barrel from such tanks has a priority. The Flowchart algorithm is'nt ready yet. When its done.... and it will take some time as client's inputs are also involved, I shall certainly post it up here so people could review it. Meanwhile, Goodluck, Gospeed, and my best wishes for you to do well. Thanks again, Raj S. Iyer

[TERdON]

Sorry for forgetting about my promise to upload a sample time-distance-diagram, I have been busy with lots of other stuff. Anyway, here is a really simplistic one. The positions are marked on the left. The line shows how the transporter robot is moving - solid line means movement with load, dotted unloaded movement. S stands for "sänk" (sink), L for "lyft" (lift). Most treatments are only fast rinsing, the pretreatment and drying last one robot cycle, and finally, the treatment in this example lasts two robot cycles, so two tanks are used alternary. The example, of course, is totally bogus. The simplest process I've seen is hot galvanizing, and even that has at least one more treatment step. Please notice the patterns that are easily recognizable: rinsing often doesn't have a minimum treatment time, so often it would be smart to put them in sequence. Other treatments have to last a whole robot cycle, and are being emptied just right before being reloaded (not greatly shown here, unfortunately). It's not necessary to do like we are, and as long as you keep your process pretty simple (hint: multiple robots and/or multiple different treatments are ways to do that easily), probably your way to solve the trouble would work as well.