Purified Terephthalic Acid

MH: Welcome, everyone to another Servomex podcast. I’m Matt Halsey, Application Development Manager here at Servomex, and I’ll be hosting once again. You’ve heard my voice a fair few times now on our recent podcasts.

Today, the topic we’re discussing is PTA, purified terephthalic acid, and I’m joined today by two of our specialists. First, we have Stephen Firth, one of our business development managers. Hi, Stephen.

SF: Good morning.

MH: Stephen also has the benefit of being an ex-head of our global systems team years ago, so he has a lot of experience in the PTA systems, which are an incredibly important part of the overall process. You’ll hear about that a lot more later. And I’m also joined today by Min-Woo Lim, who is our application development engineer for Asia Pacific. Hi, Min-Woo.

ML: Hello.

MH: And Min-Woo is actually based in Korea.

ML: That’s right.

MH: So, PTA stands for either polymer-grade or purified terephthalic acid. And this is very much a commodity chemical, so it’s used as a precursor to go into other things. We did a little bit of fact-finding – we always like to have some facts at the beginning of the podcast – about 80 to 100 million tons of this stuff is produced annually, and the vast majority is produced in Asia. The two biggest plants in the world are in Asia, and then I think that the third and fourth are elsewhere. And then we’re back to Asia again for the next five or six in the list.

So very, very dominant in Asia and China. What’s also interesting is that China makes up the largest demand for PTA as well, over 60% of global demand is just from China. And that’s because PTA is such an important precursor chemical and in making PET, especially, polyethylene terephthalate. Most people will know PET as being used in plastic drinks bottles, that’s the really famous one, and you can imagine how many billions of those are produced every year. But also in polyester, which goes into clothing.

Polyester was quite a revolution for clothing when it was discovered, because it makes things easy to wash and clean, and makes clothes very durable, so a lot of clothes have polyester in them somewhere. PTA actually manifests when it’s formed as a white powder, white crystal, with an incredibly high melting point, over 400^oC, which is why it’s so useful to use in plastics, so they don’t melt.

It’s all about oxygen safety, really. Stephen, can you give us the history of how Servomex got involved with PTA?

SF: From a historical perspective, Servomex started getting involved in PTA back in about 1987 when they were working with ICI on the Wilton site, on a then very new process to produce something called PTA to make polyesters – it’s used for making clothes and plastic bottles – and at the time, this was a new process. There’d been a few before, but that’s really where it sort of started from. And then that process, the ICI process, developed and eventually that was owned by Invista, who are now Koch Technologies. There are also other people: BP Amoco, who are now Invista, had a license, there are also licensed with Mitsui, Mitsubishi, Eastman, Tecnimont.

And, over the years, Servomex has been involved – I certainly have been involved – in working on all of these licenses; they all follow the same basic principles, same concept. The actual detail varies a little bit, but I don’t know how many we’ve worked on over the years. I did a recent presentation to a customer on the Invista/Koch license. We calculated that Servomex had been involved in the production of at least 40 PTA plants with that licensor, and quite a considerable number with other licensors.

The interesting thing is that all of them make use of a two-out-of-three voting system on the reactor, and the control of that reactor, using a highly accurate oxygen analyzer, is key to the efficiency of that process. As I understand it, they have to drive it to get the yield to get the efficiency to a certain point and then not drive it any further, which would lead to a runaway reactor and potential explosions and things of that nature.

So reactor measurement is really critical. As has been mentioned in the production of ethylene oxide and propylene oxide, it’s a really highly important measurement to do right and get right. And to a certain extent, getting the good analysis with a reasonable speed of response – typically, under 60 seconds is regarded as good response on these reactors – that gives them the optimum to control that reactor and obviously, maximize production in a safe and efficient manner.

Similarly, on the crystallizer, which is the other critical measurement, that’s really all about safety, drying, highly volatile, acetic acid off the crystals of PTA. Essentially it’s an inerting type of process, just to make sure that it never goes above the explosive limits and you get problems. And again, the oxygen level is critically important.

As I say, the licenses all roughly follow the same outline in their process, so you have the reactor, you have the heat exchangers, the separation, the crystallizers, and dryers, and they all tend to follow that same sort of basic setup. And behind that, just so we don’t forget, there are the utilities units on the plant, so there’s process heaters, and that has combustion control, which will be the Servomex SERVOTOUGH FluegasExact 2700.

Usually, it’s a pretty standard process heater. There’s emissions measurements on the process heater and on the plant itself. And then there’s usually a PSA or a nitrogen supply, or that could come from the ASU, to ensure that the crystallizers and the dryers are all maintained in safe inerting conditions – the problem being that acetic acid is, apart from being highly corrosive, also highly flammable. So you’ve got to be a little bit careful with that as well.

MH: Yeah, I think the acetic acid is, from the measurement perspective… sometimes the challenging part, if you’re taking in samples into your analyzer in your system that may have acetic acid present, sometimes you have to have some extra protection in there. So dewpoint control, of course, being very important to stop any sample condensing. I think that’s why, in the past, we have some customers that specify this gold measurement cell that we use.

SF: Yeah, I mean, it’s interesting, I was involved with a little bit of the history, I joined Servomex in the early 1990s, when ICI was working on the process, shall we say, in trying to get it to work. One of the problems they found out, very early on, is that hot acetic acid is frighteningly corrosive, it rips through stainless steel in a matter of days on the process.

And I remember one of the first sampling systems I was involved with all those years ago, we’d put Hasteloy pressure regulators at the front end, because that would obviously be resistant to the corrosion. Well, it was – it lasted about three weeks. And I remember myself, and the service engineer at the time, as a young applications guy, going to see what’s wrong with this. And we took the pressure regulator apart and there wasn’t much of it left.

So in the end, the process, the reactor, and the front end of the sampling system tends to be made out of titanium and high-grade titanium at that. And the amusing thing is that the reactors – and these are five meters wide by 25 meters tall titanium lined – you have to book your slot sort of years in advance. So once when I was fortunate enough to visit the manufacturer in Japan, they had a… I’m not sure it was an order book, it was more a promise book. Going out years when all these people were thinking about making PTA plans and they put their name down on the list to get slot for it to be built.

So I don’t know whether that’s still the case, but that was 15 years ago, something like that. But yes, the important thing is the titanium front end on the reactor itself on some of the sampling systems, washing the water, making sure you’ve got your water flowing, and cleaning out the process, drying it. But yes, there have been occasions if you put wet acetic acid through a standard stainless steel cell, it won’t survive that process. This is a plating process that’s done as a special by the Servomex transducer team. And that just provides an additional level of protection to the cell.

MH: That’d be paired with the Hastelloy, of course, the Hastelloy pipework. So I think you’re right, it’s just the level of protection. We don’t expect it to be exposed to this all the time, but if it was, it means that it would survive. Obviously, one of the big things of PTA for us is the systems element. This is shelter land, isn’t it?

SF: This is shelter… In fact the sampling system is the critical bit. The analyzer has to be good, don’t get me wrong, but the sampling system is really important. Mainly because the faster they control the reactor, the better control they have. So that means taking the tapping point as close as they can to the reactor – and one or two of the licenses actually put the tapping point on the reactor, which is great. But that means a sampling system has to take care of the acetic acid at 200 degrees and think it’s about 10 bar pressure, when it’s extremely corrosive.

So the sampling system becomes quite an integral part of the solution, especially on the reactor and the crystallizer. For the other measurements, the hydrogenation, the drying, the nitrogen lines, they’re pretty standard, there’s nothing too clever about those. But the ones for the oxidizer and the crystallizer are a critical part of the solution. This is not something that you just throw together, it’s something that’s built into at least two of the licenses, and I think three of the licenses, the actual sampling system is a key component.

So it’s really important to understand what you’re doing, how you’re doing it, why you’re doing it, and the implications of that. So, analyzer, yes, you need a good one. I agree with that. But the sampling system is really critical, especially on that reactor measurement, where you’ve got to balance your efficiency and enough power to get it going and not enough to sort of overcook the mixture.

ML: So, I wanted to know, now, about the oxidizer in the PTA process. So, this can get the gas components as nitrogen and O₂ and CO₂ and CO and acetic acid, water. If we installed the Servomex 2200, the analyzer requires dry and clean gas. So how can we remove the H₂O in the sampling system? Only by installing the moisture remover or cooling device?

SF: Yeah, basically, it doesn’t really matter which license is used. The actual detail is slightly different, but the concept for these is that you take the sample from the reactor, and Min-Woo, you’re right, there’s a lot of water and acetic acid in the sample. And it is effectively scrubbed with water. Now, how exactly that’s done there is with the license and the process. But effectively, you use water to scrub out the acetic acid and the catalyst fines and there’s a bit of bromine in there as well, normally, to clean it. And then there’s a cooling system. So effectively you have a water scrubber, and then a coolant system and you cool the sample to about, I’ll say five to 10 degrees, it’s not super critical, because once you’ve washed the sample, you’ve basically got a sample that’s oxygen, nitrogen, CO₂, CO a little bit of moisture, so it’s nice and clean. And then you can just use a reasonably conventional cooler to remove that moisture.

And then you do the sample, I mean, it’s roughly the same on the crystallizer; the crystallizer is, effectively, taking the sample and producing the flakes of PTA, the main solvent there is, as we’ve said water in the acetic acid, which is highly flammable. So again, large amounts of acetic acid and water, you water wash it and then cool it before putting it through the analyzer.

So that’s the sort of key bit, but how that is done varies with the process there, again it depends where exactly they’ve put the tapping point, depending how much water and acetic acid you get. For example, if you’re on the reactor you get quite a high volume of water in the acetic acid, if you’re doing the measurement after the cooler heat exchanger on the process, which basically takes the off-gas from the reactor and cools it down then there’s much less water and acetic acid, although there is certainly enough to cause a problem. But yeah, crudely the sample system: is water-wash the sample, dry it, do the measurement.

The beauty of the 2500 is, of course, that the cell is separated from the electronics, and the cell itself is quite a robust piece of equipment. So again, that cell is heated, it’s got the solvent-resistant O-rings, the calcium fluoride windows, and it’s separate to the electronics. So basically, the sample goes through that. And again, it’s quite reliable for measuring the carbon monoxide and the carbon dioxide, we make use of the Servomex 2550, which measures both components – it’s a multi-component, infrared analyzer – and these are mounted actually on the same sampling system next to the oxygen. But the beauty of going back to this, if you do get wet, acetic acid in the cell, it’s easy enough to clean out, and it’s no issue on the infrared. It’s quite, quite nice.

MH: We’re on analyzers, right? So the 2200, obviously one of its important capabilities is this multi-drop feature where you can have multiple transmitters to a control unit. Stephen has spoken a lot about the importance of the safety measurement on the reactors and the crystallizers for oxygen. And it’s obviously quite common for these guys to use a voting system. So, do you want to just explain what we mean by that voting system, and how it’s applied? So multiple transmitters measuring at the same point in the process…

ML: Yeah, I understand that’s why we install the redundant system for the oxygen measurement. Because, for the safety, if one analyzer supports it, and if we cannot measure oxygen, customers don’t know the oxygen value, but O₂ value is going to high, it can cause the flammable explosion. I think the customer site recommends the redundant system for the PTA system safety. Because safety is highly important for all of the process, we always recommend the redundant system for safety application.

MH: Yeah, you’re absolutely right. The safety is paramount in these processes, that’s why they’ve got this voting system. The way I always explain it to people is you have to have more than one analyzer agreeing on what the oxygen level is, or you can’t trust the reading, and you know, it gives you two benefits. You’ve got multiple analyzers agreeing with each other, otherwise the reactor shuts down for safety. Or, of course, it means that one can fail, or be under maintenance or whatever, and you have some redundancy, you have something else making a measurement. Yeah, you’re absolutely right.

ML: And also, the redundant system is better for the service team. If an analyzer fault occurs, processes just shut down. And for the service team going to the site survey, they can remove some parts because they can keep the spare parts, and they are on the site to help to the owner of the unit, they can check what is the part problem. They can check it, also redundant system, better the service side. So, we have a good point about the redundant system. Yeah.

MH: Yeah, you’re absolutely right again, yeah, it make makes maintenance a lot easier, definitely. So, here’s a question for you, gentlemen. Servomex, in terms of oxygen analysis, we provide two relatively similar products, the Oxy 1900 and the OxyExact 2200. Both Paramagnetic, both high-specification analyzers. They do have some differences, but they cover about 80% of each other, if that makes sense, in terms of where they can be used. So, why in PTA do we tend to use, or tend to find that customers want to use, the OxyExact 2200 over the 1900?

ML: Yeah, I think so, the OxyExact using the Hastelloy pipework, but the 1900 cannot apply the Hastelloy pipework. Sample components include the acetic acid. Acetic acid and H₂O meet and some problem occur in the SUS pipe. I think so we will apply the Hastelloy pipe work in analyzer, we can prevent that H₂O and acetic acid sample components. If a customer is using the SpectraExact, OxyExact, with Hastelloy pipe work, they can get the feature of analyzer useful, or prevent the sensor damage.

MH: No, very good point. So, material of construction, basically, so Hastelloy pipework. And we get asked a lot actually – I’ve been asked a lot back when I was Product Manager – why don’t you offer Hastelloy in the Oxy? Well, honestly, the main reason is the Oxy has a relatively complex machined sample block that holds the inlet and the outlet, that has some O-ring seals in it and takes the bulkhead fittings. And to be perfectly honest, that bulkhead machining is complex enough to mean that it would be very difficult to machine out of Hastelloy, it’s a very simple reason really. Whereas the 2200, the OxyExact pipework is a lot simpler, and therefore can be made out of Hastelloy tube, which is why that option is available.

So yeah, very good point about materials, the OxyExact clearly has a specification advantage over the Oxy, even though the Oxy is still very good. Now if you compare the accuracies you’re talking about less than 0.05% oxygen for the oxy but less than 0.02% for the OxyExact, so twice as good, or over twice as good, really, but still very, very good. But I suppose the oxygen measurement, you know, we’re in low percentage oxygen, typically, especially at the reactor outlet, I guess, 3 to 4% or something like that. And maybe the absolute accuracy is important enough for customers to want the slightly higher specification products.

SF: Yeah, it helps a little bit. And that’s why I think most people go for it because the OxyExact is a little bit better, the specs are a little bit better, as you say it’s 0.02% against 0.05%. And it’s surprising how many customers actually add the pressure compensation to get that little bit more. Realistically, it’s only going to add a small percentage, but there were cases where it was decided that that made all the difference. It gave them that little bit more efficiency, I guess it made them that little bit more money for a modest outcome.

But yeah, most of the PTA customers go for the 2200, the OxyExact, and it’s all about the specification. One thing that does help a little bit is when they want two milliamp outputs, which the OxyExact can do where the Oxy just has the one. I mean, obviously, you could split the 4-20. Both of them have Modbus as well. And there’s an increasing tendency to use the Modbus for the maintenance systems. But I think it comes down to it’s all about the performance as much as anything else. That’s the key driver, as it is really on the ethylene oxide as well, to be perfectly honest, or on propylene oxide, it’s all about having the best measurement you can get.

ML: One more thing, 2200 can be using the voting system, but 1900 cannot. Customer considering the price and they require the redundant system for the voting system. If they apply to 1900, they can purchase two 1900 units. But 2200 can be voting system, one display, one control unit, and three transmitter units, can install it in this analyzer system.

MH: Yeah, yeah, you’re absolutely, right, yes. Very good point. Going back to our multi-transmitter capability. So, kind of an inherent voting system rather than one that you would have to set up in a control system.

Stephen, I’m always interested in the system side, and obviously you’ve got a lot of experience with systems and you’ve touched on the importance of the systems. But this is the process where Servomex are very good at producing these analyzer shelters for the PTA process. So what would a typical analyzer shelter for PTA look like? And what would it have in it?

SF: Good question. Typically, depending exactly on the process, it has the oxygen analyzers and the CO/CO₂ and the CO₂ analyzers for the oxidizer and the crystallizer. So, what you find is that the analyzer out will be, it depends, 10 meters long, three meters by 10 meters, so they’re quite big. And they have the analyzers on panels with the calibration set up, and everything else is on the outside or scattered around the plant.

The sample is conditioned, the pressure is reduced at the tapping point, the water scrubber is as close to the tapping point as they can get it, typically a few meters, and then it’s dropped down to ground, to the analyzer house, which is sitting on the ground. Originally, there were other analyzers in there, but essentially, it’s now just the crystallizer oxygen analyzers, and the crystallizer CO₂ analyzer, and then the reactor analyzers, oxygen and infrared on panels. And then within that, you’ve obviously got the safety systems for the analyzer house – so gas detectors, alarms, HVAC system, which might be a duplicate HVAC system, because they tend to decide that the inside of the analyzer house is Zone 1, which then gets into this point that the oxygen analyzers tend to be Zone 1 analyzers. So even if the analyzer house alarms go off, and the HVAC fails, the oxygen analyzers can keep working, they don’t need the HVAC system to actually function, which is an important point, which is why the ambient temperature spread on the 2200 oxygen analyzer is really important.

It basically means it could just sit outside if it really had to, but they tend to put them in analyzer shelters, so it’s easier to look after everything. It gives the service guys somewhere to work, the analyzer technician, and they can monitor the calibration and setup of these particular analyzers.

Another thing that’s become important in recent years is SIL and the SIL setup. While this is not specifically involved with the analyzers, it’s important that the oxygen analyzers are SIL approved, because they’re included in the safety loops. And the plant will do their SIL analysis on those. But having the SIL 2 for the oxygen analyzers makes that process a lot easier for the plant. Again, what you find now is that within the analyzer house, you have one set of milliamp outputs for DCS, for running the plant, and a second set will run to the electronic shutdown system (ESD) for safety. So again, you end up with duplicate sets of junction boxes, and things of that nature. So yeah, it’s very much a standard analyzer shelter, with the key analyzers protecting from the environment where they can be maintained regularly, and in relative comfort, by the plant engineers,

MH: I was going to ask that actually, you know: “why a shelter?” Because it’s not necessarily common across all of these types of applications, but PTA do seem to really go for these analyzer shelters. But think you answered that with “somewhere to work, not only to keep everything in one place and make maintenance easy, but to give someone a nice place to work in relative comfort, especially considering where a lot of these plants are, in Asia, and we’ve got some in India, where it gets quite warm in the summer.

SF: That’s right, I mean, it is interesting. Some of the plants will go for normal enclosures, but if you try and put them on the top of the reactor, as one or two people do, then it’s quite an exposed location, it’s 35-40 meters up in the air. So to actually go and do a calibration becomes… Even if it’s an automatic calibration, you’ve still got to get your engineers up there eventually, it’s not a pleasant place to work. Whereas down in the analyzer shelter, it’s quite nice and you can do your calibrations and your checks and your regular maintenance in relative comfort. And also, you have access to all the control panels.

So you can if you like go into the analyzer house, switch an analyzer into maintenance mode. This will obviously tell the DCS and tell the control room staff that the analyzers in maintenance mode. It’s usually all documented. But it’s easier then to, once it’s been calibrated or the service complete, switch it back. And this is where the two out of three comes in that you can take one of the analyzers offline for a short period without compromising the measurement and that keeps moving on quite nicely with the other two analyzers.

MH: Thanks for listening to another Servomex podcast. We appreciate your attention. Thank you, Stephen. Thank you, Min-Woo for joining me today.

SF: Thank you, guys.

ML: Thank you. It’s a very useful time in my sight.

MH: Yeah, thanks Min-Woo. Please do check out our website, servomex.com, to find out more about our PTA offerings. On the website you’ll find our application notes, videos, and of course this podcast and previous podcasts. Thanks once again for joining today. We will see you next time.