Skip to main content
Loading Explore guides…
Process Chemistry · Acedra
Home
Explore
Chemistry
Process Chemistry
Sign in to save
Chat
If YouTube asks you to sign in, open the video on YouTube.
Watch on YouTube
Reload player
Transcript
107 segments
Copy
Hide video
0:01
you can go ahead Paul alright I'm going to present a lecture called beyond the classroom process chemistry and what I really want to focus on is the idea that anything we do in basic research ultimately gives insight into technologies which can be translated into processes and those take place in vessels so that's kind of the icon for this presentation we've selected a process vessel and that's because a process vessel is part of a larger
0:28
overall process and so the question we want to ask ourselves is how do we transfer a chemistry that we work on in the laboratory to an industrial process so chemical engineers look at what's being done in the laboratory and they say well I want to convert that to a large process facility that might cost half a billion dollars but that's that's one way of thinking about it but there's also the reverse way which is we can imagine that anything we work on in
0:53
terms of basic science will contribute to some chemistry or catalytic technology and then that can be potentially scaled up so what you want to think about is how could I select the chemistry that I work on such that it could go through a design process and ultimately become a process technology so one way to think about this is what can process analysis tell us about chemistry's to study and how can it shape our thinking about what
1:17
type of fundamental research we do now that's a very big question and I'm going to go through the process of design the red arrow right there and then I'm going to use at the end of this the the insight to get from that to think about the actual selection of the process chemistry's so the process development has both scales and times and so we start at the left with a lot of what we do and at the University where we do fundamental research we work in flasks
1:44
beakers very small scale systems and while we don't think of it as at low cost it's fitting the overall process development time it is relatively low cost and low capital you don't have to buy much equipment and the time scales are relatively short one to three years if you have an exciting chemistry that looks promising on paper you can then take that to the next scale which we would say bench scale we can do this in the laboratory the cost goes up
2:07
significantly so now we say that is moderate cost but the time scale of this would be to build a process flow system that might still fit in a laboratory and is still moderate cost low capital compared to what would be the next gift which would be the pilot plant now in this case you're building a system that might run for two to five years and now the capital investment becomes relatively significant compared to the venture or
2:31
the laboratory scale and here what you want to do is work out a lot of the challenges associated with combining chemistry's and separations together where you've invested significant money but not nearly as much as you would at the commercial scale so there's a significant problem you can find it there where the capital investment is relatively limited but ultimately a pilot process that's successful and meets all the targets would go to a
2:54
commercialization phase and this will be one to two years of construction at a lifetime we say 20 to 50 years but it could be a hundred years it could be even longer it could be indefinite a lot of process facilities run forever now in this case the cost is very high the capital is very high and the risk to start one of these without going through the bench in the pilot scale is extremely high and there are examples where people have skipped either the
3:17
bench of the pilot scale facility built a commercial scale system and completely regretted it so how do you get from from the lab scale all the way through the bench pilot scales with vertical well you need to go through the design process and here's an example of a process flow diagram this is where we take a chemistry in this case it's methanol catalytic cracking to gasoline so you can see in the process flow diagram you can see the methanol comes
3:42
in and then a bunch of aromatic compounds and olefins come out such as benzene, toluene, octane, xylenes, etc., now the way this is conducted is through what we call unit operations every reaction or separation or change in pressure or temperature occurs in a unique vessel which we call unit operation this includes reactors, pumps, compressors, heat exchangers, distillation columns, flash tanks, and some things that aren't even shown and depending on the
4:09
type of process flow diagram you look at in that case some diagrams have valves and there are just thousands of valves and other cases not the question is how do you think about all designed individually and then how do put them together this is where I want to come to the overview of the entire presentation first thing you have to know is what are the unit operations and this is where the chemistry happens so we have to be able to describe that part two
4:34
will be about once we know what the unit operations are how can we put these together in a way that makes sense in terms of the overall chemistry and number three you can imagine just looking this one option for a chemistry of methanol to gasoline on the right that there are thousands or millions of possible configurations and designs and you want to think about how do I get the one that's the best overall whatever that means
4:56
so let me start at the beginning here and the obvious place to start is with reactors because this is where the chemistry happens and it's one of the most dominant unit operations in an entire process and so how do we think about commercial flow scale reactors and we can you actually think about them quite simply here's an example of one and in this case a reactor is just a tube you have component A flowing in and it exhibits some complex velocity distribution there
5:21
might be heat transfer there might be mixing of some kind and then it comes out the other side where some of the A is being then converted to B now a system like this we think of as an extreme case because the material coming out the back side of the reactor doesn't interact with the material coming into the reactor in any way on top of that you can have some temperature change depending on the exothermicity the reaction whether we have heat transfer
5:43
equipment involved whether a mixing equipment involved that's where the the complexity comes from now the opposite version of this would be something like a tank reactor and in this case we have a mixer that as soon as A comes into the reactor all the components are mixed very quickly together in which case we have a very sudden change in the concentration and the material leaving the reactor was very recently interacting and mixing with what's
6:08
coming into the reactor so here we can have rapid changes in the concentration profiles we could have a very sudden change in the temperature from what comes in to what comes out so this is really the other extreme of what we can think of for reactor design so in terms of unit operation of reactors I can put almost any reactor on a one-dimensional scale all the way from almost zero mixing in a in a straight tube to all the way to a completely mixed system
6:32
something like a mix tank and of the thousands of different reactors for different applications out there I can really put these somewhere on this degree of mixing scale on the far left this is a very famous reactor called the riser reactor fluid for fluid catalytic cracking the reactants flow from the bottom all the way to the top in a straight tube what comes out the top doesn't really mix with what comes in and then the catalyst can be recycled in
6:54
a loop that's different than something like a fluid bed reactor here in the middle where what comes in and what goes out is sorta being mixed together in a fluid operation of either a catalyst or other solid material of some type now even more extreme mixing is something like a fermenter where everything is mixed together and at any one time any fluid element has a high probability of interacting with any other fluid element now all the elements that go into the design
7:19
take an entire course and all the aspects of mixing, heat transfer, chemistry, etc but they all fall along some sort of description of this type now once we make a product mixture we then have to separate it back out and so there's a large number of unit operations for the different types of separations that have to happen and one that's very important is just the first idea of a flash tank so material comes out of a reactor they go into a vessel
7:43
in this case 121 the light components go at the top and the heavy components come out the bottom and I can take the temperature, the pressure, and the holdup time in that vessel to do that very effectively this is most commonly used if you're doing something like a hydrogenation with hydrogen of something heavy and the liquids just readily come out the bottom a negligible vapor pressure and the hydrogen just naturally comes out the top it very quickly gets a
8:05
nice separation of heavy and light components now you can imagine that's that one stage is never enough to separate a complex mixture and so what we can do is stack many of these equilibrium stages on top of one another in which case my feed mixture will come in it'll go into a stage here where our gas and liquid vapor and liquid will equilibrate the vapor will continue to move up and at the very top it can be condensed and then some of the liquid
8:32
can be returned as reflux and come back down the column and this can continue with vapor always going up and liquid always coming back down and when I do that with multiple equilibrium stages together I can take a fraction of that condense material off the top as my distillate with high purity and I can take a fraction of the liquids coming down off the bottom as my bottoms material and as higher concentration so in this case this is
8:57
the the process of distillation you can see a couple of distillation columns here in the picture there's a large number of design aspects to this such as how much of the reflux do I need to return to the distillation column I'd like to return as little as possible since I'm I'm essentially losing that material the size of the column the amount of flow the temperatures the number of stages and the type of stages that all go together now there's been a
9:21
lot of work done in designing these the most famous simple approach to think about this is the McCabe - Thiele diagram where I can look at the equilibrium concentration between the liquid and the vapor phase as a function of the fraction and what I can actually do is figure out where my concentration of mixture comes in and then I can count off as a crude estimate of how many stages I would need to get either from the bottom to the top and use that as
9:45
the first estimate for how many stages I would need in a distillation column and from there I can go to much deeper questions about how efficient it is how many flows I need etc now there are other types of separations that are really important for process chemistry you can imagine a system where I would have synthesis gas like hydrogen and carbon monoxide with a bit of something like H2S in it in which case I don't want to do distillation because I don't
10:09
want to condense all of this and it might be a very small purity I can use something which is called an absorber stripper combination where the gas will flow in the H2S will be absorbed into a liquid you'll then be carried out in this liquid to another vessel called a stripper where I flow in something like nitrogen that pulls it back out of the liquid and now I have a stream of purified H2 and CO which I could use for something like methanol synthesis and on
10:32
the right I have a waste stream which has my waste H2S and again there's many design elements or the temperature this would be liquid i would use to circulate the temperatures the pressures etc so there are many other ancillary devices pumps are a big part of these types of facilities and they come in many there are different types there are hundreds of types and many different brands and I'll just show you three very different types this is a centrifugal pump you see over
10:56
here on the right you have a spinning head here which imparts fluid head through velocity which then converts to pressure coming out of the volume right here a positive displacement pump is very different this is a gear pump where I fluid come in its then forced around by the gears and then comes out the top these are used for different applications depending on what I'm trying to achieve something more relevant to polymers would be something
11:18
like a moyno pump where I have this kind of curved blade right here that turns it's useful for very viscous mixtures now how you select a pump what size what type of pressures you're looking for it is is um it can be done using pump curves it's a much deeper discussion but these are critical to any process now something that might be very surprising in terms of a process design is actually gas compression you would think this is very similar to a pump but
11:45
in terms of cost compressors can be very expensive both in the purchasing but also in the application in terms of electricity cost and that has to do with the efficiency of inefficiency actually of compressing a gas so a lot of effort goes into compressor design in this case you can see a multi-stage compression system here on the left and we can use this very simple symbol to identify that now because the cost can be very high you can imagine if you're doing
12:09
something like high pressure hydrogenation high pressure hydrogen is very expensive to create so we're looking for any sort of efficiency gain we can get in making a high pressure hydrogen from low pressure hydrogen and one way to do this is to do multistage compression where we'll have one compressor then we'll cool our gas coming out then we'll compress again then we'll cool etc and you can imagine the most efficient way to do this would
12:33
be having an infinite number of compressors and interstage cooling but then the capital cost gets too high so we're going to come back to this idea there's a balance between the cost of buying equipment and operating equipment and there has to be a optimum number of units in a row that would be effective but a lot of times you see these in stages of two three four compressors in a row and after that it gets to be too much equipment
12:56
okay now another part just to include this is that there are just thousands of valves any process you go into these don't add too significantly to the cost but they do add up and they do add a level of complexity and any any person that goes into work in a process system will see just hundreds of different types of valves and they all have a specific application depending on what they're trying to achieve okay so this brings us to the point now we have an idea of a
13:20
small number of process unit operations that we can use to achieve different things whether we want to have a reaction separate condensable things or separate gases etc so we've come to the question of how we synthesize the actual process there's a very simple question how do we combine the unit operations together which unit operations do we select how are the unit operations designed how do we select the order of the unit operations what are the
13:45
combined unit operations such that they are optimized and then of course when I keep saying what's better or worse the questions becomes what metrics we even use to determine whether something is better or ultimately whether of a set of possible designs something is optimal these are very big questions and I'm going to address them coming up okay so the question becomes how do we synthesize a process how do we put these different units together and I'll start
14:16
with an example here if you have a mixture of four compounds let's say is very simple mixture n-hexane, n-heptane, n-octane, and n-nonane and you want to make pure streams of each of these how would you do it but you can imagine that the mixture might come into the first distillation column vessel one twenty one and you can split this where hexane and heptane went up the distillate and octane and nonane went out the bottoms here then each of those went to their
14:40
own distillation column where we then had multiple trays and separations and we had pure streams of n-octane and n- nonane, heptane, hexane now in this case you can imagine all the decisions you would have to make you have to design each distillation column for that particular split so temperatures and pressures, types of trays, condensation reboiling, etc but there are even bigger questions which you have to address which have to do with how we put the
15:04
unit's together and so for a system like this you can imagine three options you actually more options you can imagine but I'll just pick these three you can imagine the first option which I just showed you on the previous slide in this simplified diagram or you can imagine where I always take the products off the the bottoms the first two or I could have option three where I always take the products off the distillate except for the last column so the question is
15:26
what I allude to earlier how do you know which process is the best what's better and what metric do you use and what you don't want to have is four or five different metrics because then you have to weigh the value of one metric versus the other and so there's a very simple metric which we use which we don't commonly think of in basic research but is very common in development work which is some sort of economic metric or economic potential and so this becomes
15:49
very important to think about how do the different design considerations show up as some sort of economic potential value and the complexity here comes from the idea that some of these are one-time costs and some of these are yearly repeating costs if you take a simple distillation column for this to run a reboiler you're going to need steam to run a condenser you're going to need cooling water and for any pumps or valves or compressors or anything like
16:12
that you're going to need electricity and those are going to be continuously applied to each month of production but there will be capital costs such as pumps, condensers, boilers, distillation columns, a lot of steel and equipment which you'll only purchase at the start of the project so how do you take all this into account and compare one system versus the other you can't compare just the operating costs or just the capital costs on their own and so what we have
16:36
to think of is some sometime related economic potential and the way this has been done for a long period of time is to consider some sort of lifetime for whatever you're going to do and we show this on what's called the cash flow diagram where we take into account all costs together over the life of the project and the positive cash flows are up and negative cash flows are down now we're going to put in a distillation column this is not generating any revenue so all of
17:00
these are negative costs the one in red here is our initial capital cost when we buy everything and install it and then every year the cost in blue are the operating cost of electricity steam and cooling water and we can pick some reasonable length of a project maybe 20 or 25 years that we can account this now the other problem is that each of these costs come at different times so I can't compare across the five years with a cost at fifteen years so I need
17:27
some way to compare the costs over time and this is done through a very simple idea called discounting where I take the equation you see here on the bottom right and I take the flow cash flow at any one time divided by the interest rate to the power of n n being the number of years the question becomes what rate do I use and most companies have the idea of a minimum annual rate of return depending on the type of company or if you're a growth company
17:50
like Dow BASF or Exxon your your your rate is going to be much higher because your your projects are expected to return more than a company that doesn't consider themselves a growth company might be seven or eight percent and you can then use that equation to bring all costs back to time zero so when I discount cash flows later in time are worth less than cash flows earlier in time and now all the cash flows are equal to the same amount of time which
18:17
is time zero so I can just at that point add them up so if I take all the blue cash flows and bring them back to time zero and add them up I get the purple bar which we call the present value and we sum them up we call it the net present value or NPV now what's great about this is I've now taken all aspects of the design converted them to some economic potential brought them back to the exact same period of time to a single number the net present value
18:42
which I can use to compare all of my potential costs so if I was to do that for these three scenarios I could take each of these three ways of combining the unit operations together figure out the operating cost figure out the capital costs and then determine the net present value of each of these three options and what I would like to do is select the most positive net present value now each of these options here are costs for the overall system so what I
19:07
want to pick the one that loses the least amount of money and in this case option three with a net present value of only losing twelve 21 million dollars is much better than option one or option two so that allows me to select if I'm only limiting myself to optimizing this particular combination option three now that's a very simple small-scale system the question becomes what happens when I think of an entire process when I put every
19:30
possible of unit operation together and utilize it my yearly cash flows better look something like the following at time zero I invest in my entire process that's the initial capital investment and I dig into a big hole here by putting all of the investors money in this negative cash stream once I then start the process I have a yearly cash flow which at the very least takes into sales minus the feedstock house which had better be very positive minus the
19:57
operational cost than the taxes so what will happen over time then is I'll make a lot of money back right away and I'll make less and less money back over time and when I'm considering that the costs at time zero so what I can imagine then is if I'm if I'm looking at this and evaluating in terms of the present value in each year I regain more and more of the money back that I invested and at some point I cross over to zero this is called the payback period in this case I
20:23
have a diagram which shows it at about three and a half years and from then on all the money I make would be a positive that's that at the end of the twenty years my NPV will hopefully be some large sum of money now the way to think about this is if this net present value if I've invested in this particular process at time zero it would be the equivalent of somebody giving me that amount of money right now but only would be over the length of the project so
20:49
what's great about this I know it's a lot of details about economics there's a lot more details that go into this but this basic idea gives us a single metric that encompasses all aspects of the design and so from an optimization perspective I now have an objective function that I can apply any variable I want to and tell whether it's an improvement or a negative okay so the question becomes and using this objective function we can come back and ask
21:12
ourselves we know if something is better or worse using this economic potential but now I actually have to make the decisions of how I'm going to put things together and I'm going by Jim Douglas's text book he says for any process you make there's somewhere between ten to two to ten to five power of 5 decisions so if you have to make ten thousand decisions for any process design that the problem becomes the problem blows up in size you can really never achieve an optimal
21:39
process and this is very clear about existing processes which every year continue to improve even though they've been around 20 or 30 years and so this gets into the idea there's some science of process design but there's also the art that you learn over time to try to eliminate possible designs which are never going to be viable as quickly as possible such as you can very quickly get into some sort of rigorous optimization approach to selecting unit
22:05
operations and combining them together but this has kind of led to a process design being somewhat philosophical how do you of maybe ten thousand different process designs how do you quickly eliminate the processes that are have just have no chance of being economical without risking the idea that you're eliminating processes that could turn out to be very good even though you don't initially see them and I'm trying you two kind of philosophies many people
22:31
have espoused these besides these two people Dale Rutted of Wisconsin and many others have looked at this but all you can think of this kind of two ways Jim Douglas at UMass Amherst talk i wrote the textbook where he talked about a hierarchical approach he says we're going to approach the synthesis by establishing a hierarchy of designs and using that hierarchy I'm not going to possibly eliminate something that would be very good even though I can't see it
22:55
initially so he starts very simple he asks the question if it's my process batch or continuous then he says what goes in verses what out at level two then he says within that I'm going to look on some recycle structure which we'll talk about in a minute and then ultimately separations how do I design that and finally I deal with energy Lanny Schmidt espoused to kind of a different approach where he said rather than outside in I'm going to look inside out and that
23:19
comes from his background of being a physical chemist and a reaction engineer where he said the reactor is the heart of the chemical process once you have the chemistry designed it dictates everything so step one should be designed the reactor then design the separation systems around the reactor and ultimately in step three optimize the subsystems around the reactor or separation systems now these are two ways of thinking and nobody really does
23:40
all of these by some sort of rest of the approach but it leads to it makes clear how complex it is to arrive at an entire chemical process so the way to begin is is the way Jim Douglas and Lanny actually both think which is where do you start you have to figure out what scale you're working at if you're over 10 million pounds of product per year you're going to be a continuous process and you want to be large because you get economy of scale
24:05
the bigger your process unit operations get the more you benefit from a surface area at the volume ratio you buy less steel for volume of processing volume which means your cost per unit volume of product or per gram a product or kilogram actually goes down so bigger processes should be more economical now I found that you can treat your process as a black box you can say I have these components going in A B and C I have some selectivity and conversion assumed maybe
24:35
from from chemistry that I've known and I products D E and F coming out from that I can start to at least calculate the economic potential based only on that chemistry that's known and presumably what you're making is more valuable than what you start with and if the economics look promising then we can begin the synthesis and optimization using our objective function of net present value and that's where I'm going to go in part three
25:04
okay so when we want to dig into the actual design and combination of unit operations we have to think about optimization so can if you consider a simple reaction now A goes to B plus C B might be our desired product and C is some undesired side product you have to think about how you would actually design and operate a reactor and you can imagine that you could operate a reactor at high or low conversion depending on how big it is how fast the flow rates
25:30
are what temperature etc but the I reactor does not exist in a bubble you have to be combined with some sort of separation system so when A B and C the unreacted A and the product B and by product C come out of the reactor we have to separate in some way and we're going to want to separate out our target product B from C so that's a separation cost but we're also going to recycle A back to the start of the reactor now if we think about this in terms of the
25:57
selectivity to the desired product B versus the reactor conversion you can look at the plot in the bottom right and this is a classic figure that you'll see from most process flow reactor systems at low conversion you have higher selectivity than you have at higher conversion now this introduces the the challenge of you would like to operate a system at a high conversion because when you operate at high conversion you have less A to recycle back if you
26:24
recycle a lot of A back it increases the cost by increasing the amount the reactor size because you're moving through a lot of A and you're not doing anything with it so we're very interested in from the chemistry and catalysis side of what these reactivity selectivity plots look like and different catalysts will actually exhibit different appearance some will very quickly go to very low selectivity at just moderate conversion and some
26:50
will never achieve high selectivity but they'll maintain it out to very high conversion the difference between catalyst A and catalyst B now that's important because you want to think about the combined economic potential of putting these two systems together and you can think of this kind of individually if you think of the reactor cost on the bottom left here as the conversion goes up the reactor has to get bigger to allow for
27:15
higher conversion and the cost will go up if you think of the separation costs as the reactor conversion goes down you have to recycle more and more of your unreacted feed and the cost goes up so when you combine these two systems together you're going to reach an optimum economic system not at the lowest or the highest conversion but somewhere in between this is one of the most classic process optimization problems the combined two ended operations that do reaction and separation these will
27:45
also be dictated by the catalyst that you pick and the chemistry that you pick in this case you can see catalyst A and catalyst B achieve a different optimum at different conversions but also a completely different magnitude of NPV in the case of catalyst A where you have very high selectivity and low conversion and low selectivity at high conversion this is a very common scenario or you might operate a reactor only at 15 or 20 percent conversion and recycle a lot of
28:10
the unreacted A after you separated now I bring this up because a lot of the times the chemistry's we work with we're very interested in how they perform and the selectivity conversion scenario I showed you the generic plot on the left but here's an example of two commercial our two published papers from the last two years one is in the middle is the Baeyer- Villiger oxidation of a ketone to make a lactone and you can see here this is looking at several different
28:37
temperatures the selectivity of the lactone versus conversion and you can see a very low conversion the 15 to 20 percent you're seeing selectivity of 80 - 95 percent and as you get up to above 80% its dropped down to 60% on the right is the chemistry we've been working on in the CSP where we look at the dehydra D cyclization 3-methyl tetrahydrofuran and what got me very excited is that you can see that a lot of conventional catalysts have very low selectivity of
29:06
55% and these phosphorous diesels exhibit very high selectivity of around 90% but what's surprising is how far out it goes in the conversion system where other systems will drop their selectivity significantly especially for dying systems this one goes out very far and so from a process perspective this is a very useful way to reduce the economic costs of your particular system to operate at high conversion and high selectivity so how
29:36
do you how do you think about the reactor cost now this is a very important point for the chemistry's that we work on as the residence time gets longer our reactor conversion gets higher our costs go up that's just because we're going to have a bigger reactor so we have to have some minimum productivity in whatever chemistry we work on for it to be commercially viable and if you think a lot about conventional commodity chemical
29:59
production the residence time there is seconds to minutes it's not hours or days and so we can set a minimum weight hourly space velocity of somewhere around one gram of reactant produced per gram catalyst per hour and if we're not above that the reactors just become way too large to ever be commercializable so we're interested in both the selectivity of the catalyst but also the activity if it doesn't meet some sort of minimum requirement we've operated higher
30:24
temperatures while maintaining the selectivity just to achieve a reasonably sized reactor now this brings us to the other part of that recycled problem is how do you actually separate things effectively and this gets to the question of what type of separations we can pick whatever comes out of a reactor is very frequently we'll go to a flash tank we'll do a simple separation to remove the light components from the more condensable components but from
30:51
there we have the more challenging question of how do we do something like gas separation gas separation is generally more expensive because you're using something like membranes or pressure swing adsorption to make gas products or to recycle gases back to the reactor this gets a lot more cost effective when we get something that's condensable so if we're if we have something coming out of the bottom of a flash tank isn't condensable it's
31:11
somewhere in about the naphtha to the diesel range or a little above a little below we can condense it and in which case we can do distillation which is actually much more cost-effective this actually becomes a problem again when we get to something really heavy you can imagine separating something like fructose and glucose now neither of these are volatile and they actually can be reactive if you heat them up too much we have to do something like adsorption
31:35
and that gets expensive again so in terms of volatility we like stable molecules somewhere in the gasoline diesel range now this ultimately comes back to the problem we talked about earlier if we're doing a lot of distillation and we have complex mixtures how do we put these all together and you can imagine that for the number of components that you have it gets big very quickly if you have a two component system you only have one
32:01
column and there's only one column sequence for three components you can imagine two different scenarios but for actually four components which we talked about earlier there aren't three possibilities they're actually five the five you see here on the right that gets big very quickly if you have five components which is not uncommon there are fourteen possible sequences of distillation columns you can put together for six it goes to 42 and from
32:22
there it becomes almost unmanageable but it's very common to have mixtures of ten to twelve components in just in real chemical processes so how do we possibly approach a problem this big well what we can do is use shortcut screening methods to find column sequences which are just completely unrealistic and throw those out very quickly and very quickly on a spreadsheet get down to maybe four to five sequences of distillation columns which are somewhat manageable and there
32:50
we can use a chemical process flow sheet simulation software to do a rigorous design of something like a distillation column there are many of these available ASPEN and HYSYS are two commercial ones ProII gPROMS and many others and what these do is a rigorous simulation of a distillation column or reactor or any process vessel you want to put together now from that rigorous simulation you can then get the costs of the reactor the separation cost etc and
33:18
then calculate the NPV for each sequence and then do a comparison so much as we talked about earlier that's a big undertaking to take for a reasonable chemical process separation it can take months to complete now finally the last step is frequently the how we designed the energy and this can be a big input in the process costs per min every month so reaction and separation at this point is usually designed we know reactors we know how we're going to how we're going
33:45
to separate with certain distillation columns but you can look at any process flow sheet and see heat exchangers all over the place such as this stream right here where we're actually cooling stream 4 to stream 5 and so the question becomes we have many streams that we want to heat up and many streams we want to cool down throughout the process and so the question becomes could we pair these together in a way but we take the
34:07
heat from one stream we're trying to remove and add it to a stream we're trying to heat and so when you put your process together you can break this down into two separate columns we can imagine on a temperature scale some process stream that starts at high temperature each go to low temperature and other streams that start at low temperature and need to go to high temperature and we can partner these off like kind of like a high school dance
34:29
put them two together and then figure out what is the optimal combination to do this now there are algorithmic approaches to do this and you can then find what is considered a pretty good combination of process Willis dreams even when you're dealing with a large number of possible scenarios now there's kind of two extremes you can imagine here one is that you don't do any stream pairing with heat exchangers in which case all the streams that need to be
34:54
cooled use cooling water and all the streams that need to be heated use steam the other extreme is that you've paired all the streams together in every possible conceivable way to get the most energy savings possible now the problem is both of these scenarios have problems obviously if you don't do any stream pairing you're paying a lot of energy in both steam and cooling water forever and which is not not realistic now every time you pair a stream you not only save
35:20
you save double the energy because you say both the steam and the cooling water costs for that parent stream now the other extreme though is also a problem if you were to pair every stream just to give you an optimum energy balance it would be unmanageable from a process side because you have to make sure all the streams were able to get from one point to another without affecting each other in the event of a disruption so there's some combination that works out
35:39
to be both manageable but also optimal in terms of energy savings okay so at this point you put together your entire process you have your process flow sheet which you can see is a diagram you know all the details in each unit operation like a distillation column and you know the economics of the entire process on a cash flow diagram and what you can do now is try to overtime this is comes apart from years of analysis as evaluate each of the
36:06
system components for their sensitivity to the overall economics so what I can do is I can look at a process flow sheet like this and I can pick out a few places that my I suspect are very important to the economics and I could pick four variables and each variable then I could vary plus or minus 5% say and then figure out how much changing that variable 5% up or down affects the NPV that will then tell me how sensitive my system is economically to that
36:34
variable and you can see here from the diagram on the bottom left variable one is very sensitive economically at variable two less so but still important at variables three and four seem to have very little impact on the economics I can then use that to go in and play with the variables that are sensitive and over time improve the process more incrementally so there are other aspects that play into this and I don't want to go into this too far but they're there
36:59
the other issues here that we're making a projection over maybe 20 years and at least a payback period of a couple years maybe as long as five years and other things that play into this are market forecast which are really beyond our control for example in the in the chemistry to go from three methyl THF to isoprene which I talked about earlier there's volatility both in the sales price of isoprene but also in the purchase price of something like three
37:21
methyl THF and so what you can do is actually from your cash flow diagram is generate a mix of all possible scenarios based on the prices of isoprene the product and the feedstock and look at what the possible scenarios would play out in terms of what makes money and what loses money you can even then do a Monte Carlo simulation of all the probability distributions of prices to look at how likely it is that you would lose money and this can go into a
37:47
whole field of forecasting and economics but what would you like to do over time is start with a process that's not very efficient here's an example of a process of work on where we take glucose and we make paraxylene where we started out we looked at a simple catalytic system that was way too expensive as almost $4,000 per ton whereas petrol px paraxylene can cost anywhere from $1000 to $1500 per ton and over the years as we improve the
38:12
catalyst we improve we then upgrade our economic evaluation and the costs come down as our catalyst improved to the point where our last update we did this before we put in our next best catalyst we were now approaching the point where we could potentially can compete with petroleum-based para xylene from a biomass based feedstock now let me come back to the point I made at the very beginning and summarize and conclude the presentation which is when
38:39
we started this we said that you can take a chemistry from the lab to a process application and think about what it takes to get to the process but you can also think backwards which is if I'm working on the chemistry and I'm doing some sort of fundamental science what should I focus on and what is the impact of that for potential commercialization and I've put together a couple of slides here of things that I think are key points that we can take away to think
39:01
about in our research number one that processes should be three chemistry steps or fewer now this is not a hard and fast rule but it's a simple way to think about how can I get from A to B as a product reasonably in something that could be commercialized and you can think of this in terms of use the overall yield if each step if I'm able to get both the reactor and the separator for that step to be greater than ninety percent in yield which means
39:26
a very selective reactor in a very good separation system and I cube that for each step I'm getting an overall yield of just over 70 percent now for a commodity chemical to buy and sell that means I've thrown away a little over a quarter of the cost of the money that I spent just buying the feedstock which is significant and the way people think of this is the figure on the bottom right I'm selling something a product price higher than what I'm buying it at which
39:51
is the feed stock price in green I'm losing some of my carbon in orange which are my selectivity losses I'm also paying to convert it through process costs and what's left is my margin between my production cost and what I sell it for and the problem being that that these prices can fluctuate which I said in the earlier slide but really the difference between the feedstock cost and the product price gets driven down over time as products as processes
40:17
become mature and it gets harder and harder to be competitive between the two and so what we need to do is reduce the process cost in purple and really minimize the selectivity losses so this 90 percent number is kind of a rule of thumb and three steps seems to be a good rule of thumb another thing that's very important and this comes into the feedstock cost if we think about what feedstocks are actually cost competitive it turns out that it's very challenging
40:44
depicts to find reactants that we can use that will be cost effective because those have to be also manufactured and it turns out that in a lot of the chemical industry only a few feedstocks dominate everything and that's mostly ethylene, propylene, and aromatics such as benzene now even with those that the co reactants are also extremely limited we can use hydrogen for reduction that's the overwhelmingly dominant reductant that we use in commercial systems
41:08
chlorine and bromine our commercial core reactants for oxidation we might use oxygen or hydrogen peroxide or alkylation it might be methanol ethylene propylene and for making nitrogen containing compounds it might be HCN now this is not a limited but these are the dominant compounds because the feedstock costs are so low and anything new that we would introduce as a co reactant that's more exotic would be very difficult to commercialize now another
41:35
aspect that's really been growing in strength in the past five years well from the NSF deal we get from companies is the idea of process intensification in the first sequence you see at the top you have three three systems reaction separation reaction separation reaction separation what I would like is to have a sequence where C and E are both volatile components that can be easily separated but would be even better if I especially if I work with heavy things
42:01
like biomass molecules that have a lot more oxygen is to intensify the process by combining reactions together so if I take the red and the blue reaction like separation systems I can combine those together into a single reactor and now I have completely eliminated one entire set of capital investment and I've completely eliminated an entire separation now this can go a long way and we can we don't have to combine two steps we can go three, four, or even
42:28
five reactions all together in a single reactor system and what's even happening now is a big focus areas reaction separation there's an area of research that focusing on what's called reactive distillation for actually several decades and there's many variations on this such as membrane reactors where we can do reaction separation all put together so let me kind of end with with how we think about introducing new technologies which is what we want to do
42:54
if we have new chemistry's an existing chemical process and we kind of put it up against a new chemical process it's very daunting because you can imagine if you look at the pictures one is sitting there operating right now and another one we say is a green field opportunity we could go in there pick it up and put in cement and then steel and etc for the top system existing system the capital investment is already paid off 20 or 30 years ago and so anything we put in the
43:22
ground has to have a huge capital investment on top of the operating cost so it's not really a fair competition in that regard for the touch system on the top a lot of the operational challenges have been worked out for example if the separation has an impurity that we didn't anticipate that can be a huge operational problem that's already been worked out for a new process there could be huge process problems that are catastrophic that we can't anticipate
43:45
until we go through the entire bench pilot and commercialization procedure and then finally the processes the top already has a market for all of its products and it already has sales and customers lined up whereas a new chemical or a new process would even need to take away their customers or their market and you have no defined prices so somebody asks you what can be done now this sounds extremely daunting and it is it's also the reason very few
44:10
things get commercialized but what's exciting is that a truly disruptive chemical material really can overcome all of these issues it just has to be done under a multi-year program focused on commercializing these and of course you have to start with something that could actually be commercialized let me end it right there thank you
Here to help you learn
Sign in to keep a private conversation with your course tutor.
Ask me anything about the material…
Sign in to chat