Biopowered - vegetable oil and biodiesel forum

Biodiesel => Chemistry and process => Topic started by: Drum on April 29, 2015, 05:00:44 PM

Title: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Drum on April 29, 2015, 05:00:44 PM
For quite some time now I have been intrigued by the challenge of trying to link the results of the Jan Warnqvist test to the conversion composition of the biodiesel from which the test sample was taken. Arguments have been put forward and refuted by various contributors but nobody has really been able to be quantitative in a meaningful way. Recently I rekindled some skills from a former life, so to speak, when I was involved professionally in mathematical modelling. I have developed a basic mathematical model to simulate the chemistry of the biodiesel reaction and I am happy with provisional results. I am not a chemist! This is not a predictive deterministic model in the sense that if you input known quantities for all your variables such as temperature, processor dynamics, quantities of methanol and caustic etc. etc. this will predict what is going to happen. That is simply not possible and anyhow we all know in our own way what works for us and we can all make good biodiesel. This model is a reactive model in the sense that if you perform the Warnqvist methanol test and measure the dropout then it can tell you the mass distribution of glycerides in the biodiesel that you took the sample from. To paraphrase, if you get xml of dropout in a Warnqvist methanol test you can use the model to determine what the most likely mass distribution of triglycerides, diglycerides and monoglycerides is in your biodiesel.     

Now this is where I need some help! I need to make the best educated estimate I can (or use actual experimental data if some exists) as to the solubility of monoglycerides and diglycerides in methanol. Specifically I need to make this estimate with some agreed conditions so that I can standardise the approach: the nature of the feedstock oil (soy,rape,etc), the purity of methanol, the temperature of the test and how long we wait to determine dropout amount are all things which need to be assumed. I am calling out to the extensive pool of knowledge and experience that exists on this forum to make this estimate, primarily in relation to the methanol solubility of monoglycerides and diglycerides of the feedstock oils: soybean and rapeseed/canola.
Just so I could run the program and get some feel for results I made an initial guess that the Warnqvist test dropout would contain all the triglycerides present in the biodiesel sample that went into the test, 2/3 of the diglycerides present in the biodiesel sample and 1/3 of the monoglycerides present in the biodiesel sample. I made this guess based on a crude polarity argument that the diglycerides are sufficiently non-polar that most, 2/3, of them will fall out of the methanol and that the monoglycerides are sufficiently polar that only 1/3 of them will fall out of the methanol. I am assuming all triglyceride molecules will fall out and all the FAME molecules will end up in the methanol. (Opinions on this very welcome)

 Now to some actual results. What the first series of graphs, (Figures 1-3) represent I will now explain. The horizontal axis variable represents the % dropout from the sample of biodiesel used in the Warnqvist test. It ranges from 0% representing no dropout at all (complete conversion) to 100% in which the entire sample drops out of the methanol (zero conversion/pure triglyceride oil). These are the theoretical endpoints. We then have a continuum of possibilities in between ‘mapping’ all possible degrees of conversion. The vertical axis represents the -glyceride % distribution (by mass) in the biodiesel in your processor from which the test sample was taken. As a purely numerical example suppose someone took a sample of biodiesel from their processor and after letting the glycerol settle, they measured out a 25ml sample and carried out the methanol test on it with 225ml of methanol. Suppose after the requisite time they found that they had 16ml of dropout in the flask. That amounts to 64% dropout. Looking at Figure 1 tells them that the biodiesel in the processor at the time the sample was withdrawn was comprised of (reading approximately from Figure 1) 5.5% monoglycerides, 24.25% diglycerides, 46% triglyceride and 24.25% FAME.

Figure 2 and Figure 3 simply zoom in a little closer. Figure 2 shows detail for dropout from 0% to 10%. Figure 3 is more interesting because in it we are down at the levels where most producers are aiming. I have inserted horizontal lines on the graphs indicating the EN14214 glyceride limits: 0.7% monoglyceride and 0.2% di- and triglyceride. According to my model and the assumed feedstock oil (1, 2/3, 1/3) we can see it predicts the maximum dropout percentage which can be expected to potentially meet the EN14214 standards for glyceride content. It can be seen in this case the diglyceride constraint is the most restrictive and dropout of no more than 0.32% volume is allowed. In a 25/225ml test this would amount to an absolute volume of no more than 0.08ml dropout. At the moment most home brewers aim for no visible dropout and I myself am producing biodiesel with no dropout in the test so there is no better guide than that but the model potentially allows us to be quantitative and use the dropout volume to predict the precise makeup of the biodiesel at a given point in time.

The latter point relates directly to another use for the Warnqvist test that has come to the fore and this is in a two-stage method (especially for those who go the no titration route) where they process in a first stage knowing they will not get complete conversion. They perform the test and use the dropout volume to predict how much methanol/caustic chemical to add in to the second stage to bring the process to completion. This process is typically used for fairly good oil to begin with but is known to bring down chemical usage and increase yield with less soap. So Warnqvist’s test is proving very useful at honing the economics of the process not just quality testing in the latter stages. This is one of the reasons I wanted to attempt to map the dropout % to -glyceride distribution in the biodiesel right across the range of conversion levels and brings me to presenting a couple more graphs for your consideration. They are not earth shattering revelations but suggest that my model gives an illustration of how and why the above two-stage biodiesel processing approach works. Many brewers assume, as a rule of thumb in the two-stage process, “perfect correlation” between % dropout and the amount of oil left to convert. As they are unable to determine the mix of –glycerides in the dropout the assumption is that it is all triglyceride. So the perfect correlation assumption gives them a useful upper bound on the true level of triglyceride in the biodiesel and as such means that when they do their calculations for chemicals they are always slightly overdosing resulting in very good conversion, perhaps at the cost of slightly reduced yield and increased soap production as a result of overdosing. The graphs below may, upon further development and testing of the model, allow producers to hone the chemical quantity calculation more precisely by being able to make a more informed assertion as to the -glyceride distribution of the oil remaining to be converted.

In any event, I enjoyed developing this model and graphing some results and even if this never proves to have a definite practical use I did enjoy the fact that it gives us insight and helps shed light on why certain assumptions being made by the home brewers actually work.

Figures 4-6, illustrate the deviation from perfect correlation of the true level of triglyceride in the biodiesel and the presumed amount based on the % dropout in the test. Figure 4 shows the connection right across the dropout range from 0% to 100%. It can be seen that the vertical gap between the assumption of perfect correlation and true level of triglyceride in the biodiesel varies as a function of the dropout: being better (in absolute terms but not relative terms) towards the endpoints and worse (in absolute terms but not relative terms) in the middle ranges. We can also identify the depth of the vertical gap as being a measure of the amount of methanol insoluble glycerides (di and mono) in the biodiesel. Figure 5 and Figure 6 simply zoom in for greater clarity and detail at lower levels of % dropout.

(http://i274.photobucket.com/albums/jj265/themutchkin/Correlation%20Warnqvist%20dropout/Figure%201.jpg)

(http://i274.photobucket.com/albums/jj265/themutchkin/Correlation%20Warnqvist%20dropout/Figure%202.jpg)
(http://i274.photobucket.com/albums/jj265/themutchkin/Correlation%20Warnqvist%20dropout/Figure%203.jpg)
(http://i274.photobucket.com/albums/jj265/themutchkin/Correlation%20Warnqvist%20dropout/Figure%204.png)
(http://i274.photobucket.com/albums/jj265/themutchkin/Correlation%20Warnqvist%20dropout/Figure%205.png)


(http://i274.photobucket.com/albums/jj265/themutchkin/Correlation%20Warnqvist%20dropout/Figure%206.png) (http://s274.photobucket.com/user/themutchkin/media/Correlation%20Warnqvist%20dropout/Figure%206.png.html)
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: dgs on April 29, 2015, 07:19:43 PM
Hi Drum and welcome to the forum.

Goodness me, someone has been busy doing that. I will have to read it several times to get my head around it. From what I can see some of the results are very similar to my conversion chart, but yours go into the conversion a lot deeper.

Just as a bit of background, Drum used to be a regular in Infopop before a certain person drove us all away. Nice to see you posting on here.
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Rotary-Motion on April 29, 2015, 07:54:27 PM
welcome drum, nice to see new members, as like dgs that blew my head, but in a good way

 :o
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: greasemonkey on April 29, 2015, 07:58:31 PM
Aye, welcome.
My mind collapsed after paragraph two.......
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Julian on April 29, 2015, 08:45:32 PM
Hi Drum ... welcome to the wiki and forum.

Great to see new ideas popping up, we like to try and keep a collectively open and receptive mind to innovation on here.  Like others it will take me several reads to just start to understand most of your post, but I'll persevere!  Over the years I've heard several people claim that discoloration without dropout is caused by glycerides, but no one can say definitively where this information came from.  I think Chineese whispers may have had a lot to do with it, hence quoting Jan's original email detailing the test on the wiki.  Any additional info in this area would be most interesting.

Dgs, I started ploughing through an 11 page Infopop thread on Caustic stripping that you flagged up and saw mention of Trolls etc.  What exactly was the issue ... posts by the offending individual seem to have been removed.
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: dgs on April 29, 2015, 08:53:42 PM
Julian,
I will start a new thread to explain so everyone can see. Simply so this one doesn't go off topic. Dave.
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Tony on April 29, 2015, 09:14:17 PM
That's quite a first post :)  Welcome!

Quantifying the 3/27 is quite a task thanks to the many variables.

You mention getting results from different feedstocks, I think this touches on something that I have pondered before.

As we know various oils (soy, rapeseed, palm etc) comprise a spectrum of various fatty acid chains joined into the triglyceride molecule.  What I don't know (I'm not a chemist) is how the hydrocarbon tail of each fatty acid type affects solubility in methanol when in the form of a mono or diglyceride.  If at all?

I suspect without fractional distillation to split the blend of fatty acids into groups this is something we couldn't find out - a little beyond the homebrewer - but perhaps something that organic chemists here may understand?

The best case scenario is that all the fatty acid types behave the same, but knowing what organic chemistry is like, life is never that easy...
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Drum on April 29, 2015, 09:27:31 PM
Guys thank you very much for the welcome. It probably will take a couple of readings of the post to make sense of it. And that is of course assuming I have managed to explain it as I intended. I actually developed the model and got the results shortly after Christmas but with work pressures it ended up on the shelf. I am only getting around to putting it on the forum now. When I began to put my post together today I had to think things through all over again myself, so you will be forgiven for needing a couple of goes to make sense of what I have written.

I am looking forward to your thoughts and opinions because most of you have more knowledge and experience of biodiesel/chemistry than I do. I am very glad to be able to contribute to this forum as I learned a great deal in the beginning from the wiki and the forum. Thanks!

Also if this thread gives you a headache blame Dave Smith. It is his fault I am hooked on biodiesel! ;-)
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Carrington on April 29, 2015, 10:26:49 PM
Hi drum
What a fantastic first post ,thoroughly enjoyed reading will look forward to going over it again tomorrow.

Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Jamesrl on April 29, 2015, 10:27:08 PM
Hi and welcome from another former Info pop regular.

A great opening post, l've read it and almost understand it but I'll need a little more studying of the graphs.

One question though and probably irrelevant but would the catalyst used have any bearing on your results?
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Tony on April 29, 2015, 11:29:35 PM
I am looking forward to your thoughts and opinions because most of you have more knowledge and experience of biodiesel/chemistry than I do.

I think you are overestimating us, we have a very heuristic approach to making biodiesel. ;)

However, it is very fair to say that we're interested in the scientific detail behind it!
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Twenty4Seven on April 30, 2015, 08:31:02 AM

......we have a very heuristic approach to making biodiesel. ;)


Personally, I use a top-down approach - use the good oil from the top in times of plenty, and the crap stuff from the bottom when it's scarce. :)

Very interesting article, well presented Drum.

Nick
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Drum on April 30, 2015, 10:02:27 AM
Hi and welcome from another former Info pop regular.

A great opening post, l've read it and almost understand it but I'll need a little more studying of the graphs.

One question though and probably irrelevant but would the catalyst used have any bearing on your results?


Jamesrl – not irrelevant at all because we all know the impact of too much, too little catalyst. However the best way I can answer this is to say that the model is insensitive to precise detail on catalyst, methanol, temperature, processing time, pump/mixing capacity, etc., provided none of these variables are at levels which will seriously impact the reaction and invalidate the basic assumptions of the model. To put it another way, if a brewer knows how make good biodiesel, then the model is friend not foe. The model does not attempt to account for precision in these input variables and then say “here’s what’s going to happen” as a result. Rather it looks at the Warnqvist dropout and says “here’s what happened”

I appreciate these questions because this is a mathematical model put together by someone who is not a chemist, so it will be the case that I need help is determining the validity of the model from a chemistry point of view. I could go into detail about the mathematical mechanics of the model (it is not overly sophisiticated) but I deicded not to do that initially as I felt it would take away from the main aim which was to present the results and get people's initial impression and opinions.
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Drum on April 30, 2015, 10:20:12 AM

As we know various oils (soy, rapeseed, palm etc) comprise a spectrum of various fatty acid chains joined into the triglyceride molecule.  What I don't know (I'm not a chemist) is how the hydrocarbon tail of each fatty acid type affects solubility in methanol when in the form of a mono or diglyceride.  If at all?

It will affect the solubility you are spot on, certainly in any informatiion I have read on this. We can never hope to make this completely prescriptive, because there is a distribution of fatty acide tails, but we may be able to do a bit of 'averaging' for specific oils or known mixtures of feedstock oils.

Quote
The best case scenario is that all the fatty acid types behave the same, but knowing what organic chemistry is like, life is never that easy...

Yes we could do with some lab results on this (though I am informed by one chemist that there does not appear to be rersults published on this precise aspect of methanol solubility of monos and dis for specific oils), either that or we can do a bit of trial and error with our assumptions on the methanol solubility of these see if it fits with our actual processing experience. dgs has indicated already that his result align at an initial reading.
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Drum on April 30, 2015, 10:21:25 AM

Personally, I use a top-down approach - use the good oil from the top in times of plenty, and the crap stuff from the bottom when it's scarce. :)


Me too!
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Drum on April 30, 2015, 10:30:41 AM
Over the years I've heard several people claim that discoloration without dropout is caused by glycerides, but no one can say definitively where this information came from........

..... Any additional info in this area would be most interesting.


Julian - I have had it confirmed by a chemist's opinion also that there will be some solubility in the mono and diglycerides but as outlined below and as you indicate yourself, there does not seem to be any real difinitive evidence. Perhaps becasuse of the complexity Tony alluded to in the length and distribution of fatty acid tails and the unlikely existance of a one-size fits all rule for. The length of tails and the degree of saturation/unsaturation will impact but precisely how is the $64m question.....
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: dgs on April 30, 2015, 11:27:16 AM
Someone once said that the process mixing can be too efficient (vigorous)
I have found that since I have changed to powerful pumps and eductors there is glycerol in suspension with the bio. I assume because of the mixing on a molecular level. So much so that it is impossible to do an accurate soap test because of all the glycerol in suspension.

I am not concerned about this as de-mething/water washing removes it.

On the mono's/ di's subject, if indeed most of the di's are detectable from the 10/90 test then most can be removed by having a clear pass. As I have found when vigorously water washing the mono's can also be removed.

The more we know about biodiesel production, the more questions there are to answer.
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Drum on May 08, 2015, 11:04:12 PM

The more we know about biodiesel production, the more questions there are to answer.

Yes Dave, that is the nature of science and is what makes it all the more intriguing.
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Drum on May 08, 2015, 11:22:46 PM
Hi folks, I was lucky enough to be able to do a little bit of promotion for the forum. When I first produced the results I sent them to Jan Warnqvist himself. Jan very kindly read my description of the basis for how the model works. He was positive and complimentary of the model but was able to point out the difficulties that will exist in getting data on the solubility of diglycerides and monoglycerides in methanol, even for any specific oil, let alone a range of oils or mixtures of feedstock oils. However it is certainly not impossible, it just appears there is no specific research/data published to answer this question at present. I have had some useful feedback from Jan recently and he has just posted a link to this thread/forum on the LinkedIn professional network.

Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Tony on May 10, 2015, 04:24:28 PM
Hi Drum, you've prompted us to create a LinkedIn group for biopowered, https://www.linkedin.com/grp/home?gid=8304445 - though really I confess to knowing little about LinkedIn and whether this is advantageous in any way!
Title: Re: Linking Warnqvist dropout to glyceride distribution of biodiesel
Post by: Drum on May 10, 2015, 05:30:31 PM
Tony,

I cannot see LinkedIn doing the forum any harm whatsoever, even if it does no perceivable good. The advantage I saw in what Jan is doing is associating his name, which as we all know is well respected in biodiesel circles, with the biopowered forum and what I hope is that people will visit the forum to read the thread, because he linked directly to the forum, and stay as a result. I will certainly join your group on LinkedIn but I agree that the preference would be to draw people from LinkedIn to here. But the group on LinkedIn will do no harm in my opinion.