#52: Tracking and Tweaking Your Extraction | Dr. Marco Wellinger | Expo Lectures 2018

Today’s lecture explores the different chemical and physical markers that characterize espresso extraction. By using TDS, acidity, caffeine and chlorogenic acids extractions of different grinders, machines and brew ratios can be compared to each other. Recent studies by Dr. Marco Wellinger’s lab have demonstrated that for a given roasted coffee TDS, as measured by a refractometer, correlates well with the content of a range of components in the coffee brew, such as caffeine, chlorogenic acids, and titratable acidity.

Dr. Marco Wellinger is a Q Arabica Grader and research fellow in the group of Chahan Yeretzian at the Institute of Chemistry and Biotechnology at ZHAW Wädenswil in the field of chemistry, technology, and sensory analysis. His fields of research include instrumental analysis of volatile aroma compounds from coffee (gas chromatography and mass spectrometry), espresso machine and grinder technology as well as sensory analysis of coffee. He was the lead author in the heritage SCAE water chart published in 2016.

Related Links
Presentation Slides

This presentation gets very technical, so we recommend listening while looking at the presentation slides.

Table of Contents

0:00 Introduction
2:00 Introduction to the presentation and Dr. Marco Wellinger’s background
7:30 The most important markers that characterize extraction (Total solids, Total Dissolved Solids, Chlorogenic Acids, Caffeine)
11:45 How to apply chemical markers to study extraction
22:00 An exploration of what is espresso, from the point of view of the coffee industry and consumers
26:40 How to combine the brewing control chart with other
chemical markers
39:00 Conclusions of these experiments and outlook for Marco’s future experiments

Audience questions
46:30 When plotting out TDS samples by the second, what physical methods did Dr. Wellinger use to gather this data?
47:30 Why did Dr. Wellinger choose calcium carbonate to express acidity in coffee?
49:50 What type of software did Dr. Wellinger choose to express the correlation curves?
50:45 Outro

Full Episode Transcript

0:00 Introduction

Heather Ward: Hello everybody, I’m Heather Ward, SCA’s Senior Manager of Content Strategy and you’re listening to the SCA Podcast. Today’s episode is a part of our SCA Lectures series, dedicated to showcasing a curated selection of the extensive live lectures offered at SCA’s Specialty Coffee Expo and World of Coffee events. Check out the show notes for relevant links and a full transcript of today’s lecture.

As we’re taking some time to work through our 2019 lecture recordings from Expo, we thought we’d take this time to share great content from 2018 that hasn’t been released yet. If you want to see live lectures in person, visit worldofcoffee.org for a full schedule of our lecture series in Berlin this June.

Today’s lecture explores the different chemical and physical markers that characterize espresso extraction. By using TDS, acidity, caffeine and chlorogenic acids extractions of different grinders, machines and brew ratios can be compared to each other. Recent studies by Dr. Marco Wellinger’s lab have demonstrated that for a given roasted coffee TDS, as measured by a refractometer, correlates well with the content of a range of components in the coffee brew, such as caffeine, chlorogenic acids, and titratable acidity.

Dr. Marco Wellinger is a Q Arabica Grader and research fellow in the group of Chahan Yeretzian at the Institute of Chemistry and Biotechnology at ZHAW Wädenswil in the field of chemistry, technology, and sensory analysis. His fields of research are instrumental analysis of volatile aroma compounds from coffee (gas chromatography and mass spectrometry), espresso machine and grinder technology as well as sensory analysis of coffee. He was the lead author in the heritage SCAE water chart published in 2016.

This presentation gets very technical, so we recommend listening while looking at the presentation slides. There’s a link to them in the episode description. I’ll jump in occasionally to help you follow along.


2:00 Introduction to the presentation and Dr. Marco Wellinger’s background

Dr. Marco Wellinger: Hello everybody. First off, I’d like to thank the organizers for giving me the opportunity to give a talk here. It’s actually a talk I’ve also given back in Europe at least twice and hopefully you’ll all enjoy it. The title is called as you can see tracking and tweaking your extraction. Honestly, it’s more about the tracking part up till now. My name is Marco Wellinger. I work at the Zurich University of Applied Sciences in Switzerland and first up, a little bit about me, why I’m standing here in the first place. I started out about 12 years ago with enjoying espresso and I got into it. With preparing espresso I got different machines. For example, like this small touch of classic where I even bothered to put in a PID even though it has such a small boiler that the temperature drops by five degrees or like ten Fahrenheit during the extraction anyway, I also yeah, but at least I got a reproducible starting point.

I also got the like a manual lever Olympic Cremina and my workhorse is the home is usually the GS3 you can see there. Then about six years ago I got into coffee research and what we do there is all sorts of research relating to coffee. We have grand grinding machine manufacturers, coffee machine manufacturers or roasting machine manufacturers. Initially, we’re mostly specialized on the aroma analysis, but nowadays we do a lot of other stuff too, like non-volatile analysis and also sensory analysis.  So, espresso technology is complex. You have standard parameters, which are coffee freshness, particle size distribution so the grind, deepness of cup and water flow and there are a couple of newer parameters PID control has been around now also almost ten years I guess then more and more machines also offer the option to do pressure and flow rate profiling and lastly water composition has also been an issue and nowadays there are even some manufacturers who introduced real-time monitoring of the water quality that goes into the machine. But there’s I think those machines are not on the market yet, but I’m curious how it will work, it certainly a good idea to start this.

So, tracking your extraction quality indicators of for your cup. If on one side you have the machine technology, which is ever getting more complex. Sensory revelation, it’s certainly feasible, but it’s hard to conduct in an objective and reproducible manner. It just takes a lot of people and a lot of time especially if you also want to cover it. If you want to have a result that represents different target groups, like different markets with different cultural backgrounds and preferences. So, if you measure chemical and physical markers this offers you the chance to characterize an extraction independent from personal preference. So, what are the most important character markers that characterize your extraction?  I’d start off with TDS and what it is. TDS is simply put, the measure of how much coffee is in your beverage. It’s about actually dissolved solids otherwise particles that get into the beverage because it wasn’t filtered completely will be included by their full weight. So, for example, if you do a French press and you’d have like this typical sludge at the bottom of your cup all these fine particles, if you wouldn’t filter it they would account by with the full weight, even though they weren’t maybe extracted fully. They would be counted in dissolved solids.

So, classically the method to do this is vain weighing in evaporation. It’s an expensive methods and it’s very slow because it usually takes a matter of couple of hours. On the other hand, measurement by refractometry is a very fast measurement and it’s also very precise. What I want to address here also, because it’s an issue that I’ve encountered a couple of times is that there’s TDS of water and TDS of coffee and the two unfortunately go by the same name, but they use very different principles to actually measure it. TDS of water is based on or at least as it is used in the coffee community it’s based on electrical conductivity. So, here is a kind of a comparison of the two. So, TDS of beverages as I said is based on refractometry, the measurement range usually is somewhere between 0.1% up to 20% Technically you could also go higher than 20% but it’s this is quite rarely used. Only it’s a high precision method, you have less than 5% relative error, which is something like all these standards for analytical measurement devices also in our lab.


7:30 The most important markers that characterize extraction (Total solids, Total Dissolved Solids, Chlorogenic Acids, Caffeine)

Dr. Marco Wellinger: The TDS of water on the other hand as I said is based on electrical conductivity and the range is about thousand-fold lower. So, we start at ten-thousandths of a percent and you go up to 0.1% maybe. So, our PPM, which is parts per million, like percent is parts per hundred and PPM parts per million. It would be one to a thousand. This only gives you a rough estimate, typical errors lie around 30%. But it’s actually if you have this error, if your error is actually much larger. It depends a lot on the actual composition of the water. So, if you don’t know what exactly you have this is only going to be a rough estimate always…

So, back to the TDS of coffee beverages. I want to say a few words about what I addressed already like the total solids against total dissolved solids. If you take total solids then all particles present in your beverage count towards this number of total solids, but in contrast to this if you have total dissolved solids only substances that actually have been extracted and dissolved into your beverage count towards total dissolve solids and when you compare different expression methods, also the ratio between the two also varies. So, for example for filter coffee as you filter it through a very fine paper filter the two won’t be much different. That’s also why you don’t have to actually filter your sample if you take TDS measurements of a filter coffee. But, on the other hand, if you do French press or espresso where you have quite a lot of suspended solids the difference between the two can vary and it could be quite significant.

So, here I got a slide illustrating this total solids versus total dissolved solids. It’s a series of duplicate French press extractions. We conducted at 50, 60 and 70 grams per liter and as you can see by the graph the unfiltered method overestimates the extracted solids by about 7%. So, we can see that down at the bottom on the horizontal axis we have TDS filtered, so dissolved and here we have the unfiltered ones. They’re always about 7% higher. Well, of course, this will not only depend on the extraction method you choose. If it’s 7% or if it’s more or less, but it will also depend on the actual mesh of your French press. It will depend on the grind degree. The finer you grind the higher percentage of [08:27 inaudible] you have and then the more [08:30 inaudible] will also end up in your cup.  So, it’s certainly a good idea to filter them.

Next up in the chemical markers, we have chlorogenic acids and caffeine. The two are among the most abundant substances in coffee. Chlorogenic acids are related to positive health effects from coffee. We use the plural here because it’s actually not a single substance, but it’s a whole group of substances which have similar properties. Chlorogenic acids are negatively correlated with roast degree. So, light roast contain more chlorogenic acids than dark roast. Also, the faster you roast the more chlorogenic acids will be retained, the slower your roast the more they will be degraded. Caffeine, on the other hand, is almost constant across different roast levels because after the first crack when you have barely any water left in your bean the little amount of weight you lose from your bean is about the same percentage as the caffeine overall. So, mostly you’ll look you’ll end up with the same caffeine content whether you do a light roast, a medium or a dark roast.


11:45 How to apply chemical markers to study extraction

Dr. Marco Wellinger: So, now I want to show you some examples on how we actually apply these chemical markers to find something about extraction. First up we have a study where we used three different grinders and one machine, and we burned through 300 kilos of coffee and it was a performance study on grinders from fully automated machines actually for home use. We had one from Italy, one from Switzerland and one from Germany and the coffee we used was not really what you would call the specialty coffee typically because it contains between about 60% Arabica and 40% Robusta. But this is a reflects kind of the average coffee that is being sold on the mass market at least in Switzerland and I guess in a lot of parts of the world it’s still the norm that the kind of coffee for the masses still contains a significant amount of Robusta.  And then in order not to compare the extractions of the different machines with each other, but really just a grinding, we extracted the ground coffee on one single machine while the ground powder [10:58 inaudible].

So, what we did was we measured physical and chemical characteristics of the brews we produced this way over the course of grinding 100 kilos on each of the grinders. We had a first step after one kilo,25, 50, 80 and 100 kilos and we always did two different beverages. We did a lungo or coffee crema and on the other hand, we did an espresso. So, here’s the first part of data. We have 180 extractions at two different Brew ratios. I plotted them on the brewing control chart. So, since we have two different beverage types with the constant brew ratio, we have the lungo down at the bottom. You can see it here. These are all the lungo extractions. They’re extracted at a brew ratio of about 12 and had TDS in the range of o1.5 to 2.4%.

On the other hand, we had the espresso at the brew ratio four and some of you might work a bit because this is again not something that is very representative of specialty espresso cause most people in specialty they pull the espressos rather around maybe brew ratio two. I’ll get into that a bit more later on but anyway, this was for us a good compromise between what is used in specialty coffee and what is used in the mass market at brew ratio four. Okay and here we have the same again, but not really. Now, I switch the axis. We don’t have any more the TDS up here. What we actually plotted it against the acidity, but it looks pretty much the same and what we refer to as acidity here is the titratable acidity to pH 6.6 and this correlates quite well with the sensory perception of acidity and for the sake of simplicity, I’ll just refer to this as acidity on all three of slides. We prefer to use this over pH because as I said, this correlates better to the sensory impression of acidity and what you do there in a lab is basically take your coffee beverage which are usually at a pH quite low like pH 4 or pH 5.

So, they’re quite acidic and what you do there you add the opposite of an acid to it, a base or lye and you add so much lye until your pH reaches 6.6 again and then you can measure how much acids were in the cup that actually react with your tongue to give you this acidic impression. But of course, this correlation of what we measure in the lab and what is perceived on your tongue only holds up to a certain point. You’ll know that if you go to a completely over extract beverage that the acidity perception will actually decrease again, but this is not so much because you actually have less sitting in your cup but because other extracted compounds overlay the impression of acidity, compounds that are bitter or astringent.

So, if I put the two different beverage types together on one scale I get this graph and you can see that TDS correlates very nicely with the acidity of the whole range because before I always had extraction yield down here then they split up because they are differently concentrated. But if you plot the acidity against TDS you see down here we have again the lungo extractions and up here we have the espresso extractions and you can see even the outlier up here is still very close to the fitted line and what is nice here is not only is the acidity kind of a linear function of TDS, but it actually goes through 0. So, this means it’s directly proportional. If you have double the TDS in this case, there are no, it’s not double but you have a ratio of 144 times the TDS. So, if you increase your TDS down here by one increment the acidity will go up the same amount as it will go up here.

So, acidity is directly proportional to TDS because as I said, it goes through 0. If I do the same for pH you get the huge scatter as I mentioned. You see that the line doesn’t really correspond well, and this is not just because pH is a logarithmic scale and such even if you do it in the in straight concentrations, you’ll also get the same scatter.

Okay, next up we have professional machines which we studied. We compared the fully auto machine with the hybrid and the semi-auto. The hybrid is, I’m not sure if everybody is familiar with it. A hybrid is basically a machine that grinds those in temps by itself. and the only thing the user has to do is at the end of the extraction takes out the portafilter empties it and put it back in. There are a few manufacturers who came up with this concept. and the performance test was basically to find out how well or how they compare to each other and we all adjusted all the different machines just to produce an optimal beverage from a sensory point of view and we ended up, all of the extraction ended up in the optimum of 18 to 22% of extraction yield. We had one exception; one local coffee was actually a bit under extracted at 16.6%.

Getting to the results. First up I got correlation of TDS against caffeine. What I haven’t mentioned yet is that in this experiment we used two different beverage types and two different roast degrees. So, down here we have a lungo with a medium roast then up here we have three types of espressos, two of them were extracted with a darker roast and one of them was extracted with a medium roast and you see that all of them line up on the same graph. This is what I mentioned in the introduction that caffeine even though one coffee is roasted darker it was the same beans, but they were roasted darker. They still got the same correlation of caffeine to TDS and again, this line goes almost perfectly through 0 because this is a small number. So, regardless of where you are caffeine is always caffeine in milligram per gram of beverage here is always 55% of TDS which is a nice result that we saw that regardless of beverage type or even roast degree we got the same correlation.

If you go then to acidity we got the picture that we actually also expected to get because as probably everybody here knows that acidity goes down after first crack, overall acidity so the darker you roast the coffee, the less acidity will it will contain and so only the medium roasts down here from the lungo and the espresso lineup, but the dark roast they kind of drop down because at the same TDS they contain less acidity.  Again, we have quite a small intercept like where the line hits the axis is quite close to it, I mean the number 43 obviously is not really close to 0 but if you compare it to this one here, it basically means that if you are at a TDS of 0.2 or 0.3 then you’re already at the. In the city of 0 so it’s only in the range down here below 0.5% TDS. You couldn’t use it. but up here like from 2% up to 7, 8% it holds well. Then go to the chlorogenic acid. We see pretty much the same picture as we saw for the titratable acidity. As I mentioned chlorogenic acids, they get degraded with the progressing roast level and we pretty much got the same picture again. The medium roast lineup well, the dark roast drop down.



22:00 An exploration of what is espresso, from the point of view of the coffee industry and consumers

Dr. Marco Wellinger: So, next up is a short interlude about espresso and what it actually is. They’re very different points of view of what consumers expect and what the industry actually offers. Traditionally, there’s been a few definitions around for espresso. There’s the Illy book, which is now also I think something like 15 years old at least. They defined espresso beside all the parameters like pressure and water temperatures. They defined espresso as using a ground coffee dose of 6 1/2 grams plus/minus 1/2 grams to a beverage volume of 20 milliliters plus/minus 5. Then the Istituto Nazionale Espresso Italiano (INEI) has also a definition which is kind of close to the Illy one, the 7 grams plus/minus 1/2 to a beverage volume of 25 millimeters plus/minus 2 1/2. So, it’s actually a bit of a smaller range they allow for. Then I got the number from for typical specialty coffee or WBC espresso. I mean, of course everybody expects doubles there, but I kind of calculate it down for single to be comparable to the other definitions. So, ground coffee those is usually around 9 grams plus/minus 1. Beverage weight is then 18 grams plus/minus 3 and the beverage volume is about 30 milliliters.

So, if you put this into a graph it will look something like this. You see this part here which we represent the espresso brewing ratio. So, it’s the mass of the beverage divided by the mass of the coffee use. So, if you prepare a 36 gram espresso shot with 18 grams of coffee you have a brew ratio of 2. So, you can see I listed, unfortunately. I still haven’t included the newest barista champion, but you can see that most of the past barista champions were quite close to brew ratio of 2. Here I got numbers from James Hoffman. He did a poll on the what brewing ratios people used and you got something like 2,000 replies so it was quite a lot of people who replied there and the range where 90% of people who responded was around, I’m sorry actually it was shifted a bit, the error bar should be from 1 1/2 to 2 1/2. Then if you go over here then here I got the Illy definition I had before and the see here you have a huge range indicated and this is because they define the beverage based on volume including crema and then if you play around with the numbers, you take a lower dose, for example within the variations they allow for you take a low dose and a bitter higher beverage volume. You get a huge spread because you never know what exactly the crema of volume is that they assume. If you go out to specialty coffees often, you get something around 10 milliliters of crema, but it could also be 5 or 50 milliliters, and this, of course, impacts the actual brewing ratio lot.

Then INEI is a bit tighter as I said. Here we have capsules which are actually quite dilute preparations considering that they’re usually called espresso. We have brewing ratios of around 4 up to 8. Then I also plotted the typical semi-auto extraction in Switzerland which is often around Brewing ratio 5. A typical professional fully auto is usually bit lower, brewing ratio 4 and the typical fully auto at home is at brew ratio 3 but this is not because they actually pull concentrated shot. but because most fully automatic home machines are very inefficient. They often have extraction yields in the range of 15% percent and lower so 10 to 15%. So, what they do is they just use loads of coffee and then you end up with a kind of an artificially low brewing ratio, like a more concentrated brew ratio. But if you would look at the TDS that would actually not read higher than these two here.


26:40 How to combine the brewing control chart with other chemical markers

Dr. Marco Wellinger: So, next what we did is combining the brewing control chart with the chemical markers. I showed you before but before I go into more details, I want to give you a quick primer on the brewing control chart. I guess mostly everybody will be familiar with this one. This is the brewing control chart from Heritage SCAA, which has been developed for filter coffee and what you note here is that they also use the term brewing ratio but usually or traditionally that’s been used to define their coffee though so 60 gram per liter of coffee or you can do it, of course in ounces and pounds too but the important thing is that they use the brewing water and not the beverage weight. So, if you then would yeah, I guess I’ll go to next one because this makes a big difference. For espresso usually it’s not possible to determine how much brewing water you used because you still have a wet puck at the end, you have some that gets exhaust through the solenoid which makes it difficult to actually measure the amount exact amount of brewing water. So, for espresso people have naturally gone to a brew ratio, what I call it not brewing because I refer to the brew, the beverage and not the water I used to brew it. They used the ratio of coffee dose to the beverage weight. I think one of the first people who use this was Andy Schechter and what you end up with is brewing ratio often it also has been expressed in percent that you say like if you make the traditional brew ratio two I call it, they will call it 50% but I prefer the other number because this makes it easier if you go to filter coffee and you say 60 gram per liter. Actually, if you go to the traditional brew ratio that would call it a brew ratio of 16 1/2 or something like that but this is based on the brewing water but if you take the beverage then you’re at 14 1/2. But this is of course, like you having numbers like 15 or 16 is probably handier than throwing around percentages in the range of 0.15% or 0.18%. But the important thing is that by using brew ratio you can actually come up with a very easy equation to relate brew ratio and extraction which I show here. So, extraction is nothing else than the strength multiplied by brew ratio.

So, for example, if you have an espresso at a strength of 10% TDS and use the ratio of 2, then you automatically know this corresponds to an extraction yield of 20% because 10 times 2 is 20 and you can do the same for filter coffee. If you have a brew ratio of 14 and the strength of 1 1/2 then this gives you an extraction yield of 21% and I saw that some people they now also suggested to come up with a brewing control chart for filter coffee using brew ratios but what I saw so far is that they actually use the brewing water instead of the beverage mess and then I think this will inevitably lead to confusion because then you have kind of two different numbers which look kind of similar but they’re not actually the same. So, I would very much welcome if you switch to this type of ratio for filter coffee that they actually would use beverage mass and not brewing ratio. I mean, of course, it’s still very useful to have the 60 gram per liter these doses but to actually calculate or do something with it I think brew ratio defined this way is handier plus it’s the same than for espresso and for filter coffee.

Okay, if you then plug it into a graph it can look something like this. Filter coffee at the ratio of 15 gives you a TDS of around 1.2 to 1.4 in the [28:53 inaudible] Optimum or WBC espresso, typical one of the brew ratio of 2 is in a range of 8 to 12% TDS and you can see one brew ratio always represents one line, so we have to filter coffee off of brew ratio 15 down here. Like a more dilute espresso ratio 6 here then 4, 3 and brew ratio of 2 and as I said, if you are at 10% TDS brew ration 2 then you end up at 20% extraction yield. So, what are we doing with this then? What we want to do is we want to track extraction throughout over time. So, throughout not just measure the end result, but actually, measure how the extraction develops over time. Matt Perger was I think one of the first to introduce this and what you do is nothing else than visualize how TDS evolves against extraction yield over the course of your extraction. So, in a small series of extractions, we explored how TDS and other chemical markers evolved throughout extraction.

First up we have a quite an unexpected result from a fully automatic which was that we got 18.7% extraction yield in a matter of 8 seconds because often there’s always this talk about how can you extract a decent coffee in under 10 seconds and I must admit I also thought this is absolutely not possible at least for an espresso and the astonishing thing is that the extraction was so efficient. You see that the point here, this is the first point after 1 second, these 2, 3, 4, 5 so they’re always spaced in the same time, but you can see the extraction the longer it gets, the more inefficient it becomes because the spacing on the extraction field axis is getting smaller and smaller. So, between 6 and 8 seconds, there’s barely anything happening anymore. But in the first 4 seconds of the extraction, you reach almost 50% of yield. So, maybe if somebody finds out a bit more efficiently, it will actually be possible to reach a 20% extraction in 10 seconds but I’m not sure if this is actually a goal we should head to.

So, you can plot the range of paths on the brewing chart. I used here data from a lot of different extractions we had. We had a very concentrated shot starting out at more than 25% strength and depending on flow rate and even as you do extraction you can take very different paths on this model. So, back to my actual data, we did an experimental series with a fully automatic machine. We used the single origin coffee from Brazil with a medium roast double. We used a 20 second extraction time and a brew ratio of 3 and we split into 8 fractions and in order to get a more even spacing we chose to have first shorter time intervals because the extraction initially is more efficient and then at the end make the intervals longer. So, first we had 4 times 2-second intervals and then we had 4 times 3-second intervals and what we measured was TDS, titratable acidity, caffeine, and chlorogenic acids.

So, the extraction starts out to highly concentrated and the first half of the beverage contains 2/3 of the dissolved solids and then I guess I should go into the graph a bit more. We still got the brewing control chart here with TDS against extraction yield but now I introduce the second y axis and here I put the shot weight and I know okay. I ended up at 40 grams of shot weight and as I said like the first half of the beverage here, it’s 20 grams is already at around 15% extraction deal. If you look at the acidity, you see something that most of you will probably have expected that the acidity starts out at a much higher rate initially than the shot overall. That you get a lot of acidity initially but actually it doesn’t go down to zero in the last part of the of the extraction, but it actually keeps on extracting a little bit of acidity. If you look at chlorogenic acids, you get pretty much a straight line and having a straight line in this sort of plot means nothing else than the chemical compound you’re measuring, in this instance the chlorogenic acid extracted at the very same rate than the TDS overall than the overall extractable compounds.

If you look at caffeine, it’s kind of the opposite of acidity a bit. You start out kind of at a flat rate, at a low rise but you get a huge increase compared to the rest at the end of the extraction and this is basically because caffeine keeps on extracting it’s not so easily soluble and back here there’s almost no other compound still extracting a lot, but they’re still a considerable amount of caffeine coming so you don’t move in extraction yield by much like barely any but it still extracts another 10% of caffeine at the end. So, this is why it’s almost vertical then it at the end and if you then put it all together you can kind of see how the different dynamics compare to each other. That the acidity as expected is kind of the fastest extracting, chlorogenic acid is very even and the caffeine is kind of the slowest of the of the bunch.

Then actually I got also the same thing I just showed you on the time axis. I guess it could be a bit confusing now, but I just want to, for completeness sake, I want to show you also what it looks like if you look at it in time and not in relation to the extraction yield. If you do it cumulative, so I always add up what I already had you see that the caffeine already reaches 90%, acidity already reaches 90% at around 12 /13 seconds. whereas caffeine is still at 70% only and then caffeine keeps on extracting while most of the others they level off. If you look at it by fraction, so differentially how much does it change in each fraction?

You can see that the chlorogenic acid, overall TDS and the acidity they extract extremely high initially but then drop down. You can also see the shot weight here which is kind of like the flow rate here. That you have a bit of higher flow rate initially then it drops down and then increases again and the caffeine, well, I told you before is actually not perfectly constant, but it’s quite constant over time. It is a bit higher initially, but it’s in contrast to the others especially. It doesn’t drop down at the end because it’s so hard to do to solubilize it. and this is also one of the reasons why people always say per gram of coffee you get more caffeine out of filter coffee, then an espresso. This is simply because you put through a lot more water. So, caffeine has more water to dissolve in.


39:00 Conclusions of these experiments and outlook for Marco’s future experiments 

Dr. Marco Wellinger:  Okay. I’m already at my conclusions actually TDS measured by refractometry reflects expression levels very accurately for a broad range of brewing conditions in coffees. For one coffee and similar extraction levels all come markers correlate very precisely with TDS even across different beverages and machine types and using the acidity alone you can relate the city potential of the bean, the buffering capacity of your water and the resulting acidity of the coffee beverage. So, the acidity you have from your coffee get reduced by the alkalinity of your water and then it results in the acidity of your beverage and our tongues work much rather like a titration machine which measures the acidity than as a pH meter and it appears that it could be very well sufficient to measure TDS against the other parameters once, only once and then subsequently estimate chlorogenic acid, caffeine and titratable acidity only by measuring TDS alone and then using the same factor you determined before. But, of course, you have to determine this for every coffee and roast level and of course this also has its limitations. Seeing the dynamics that I showed on the last couple of slides not everything evolves at the same speed and so if you go to very high extraction yields or very low extraction yields these correlations will not hold up anymore because they don’t evolve at the same speed but if you compare extractions at like 20% plus/minus 2 then it will still be very precise.

Then I got to a shorter outlook. What are we into next? Currently, we’re about to do much of the same study we did here with coffee. We had quite an interesting result actually where we saw that the coarsely ground coffee extracted faster than the fine ground coffee which is kind of counterintuitive but what we also saw there is that it seems that finely ground coffee or we know for a fact that finely ground coffee degases faster and degassing can also be like additional resistance. Basically if you have an old coffee, it’s almost impossible to pull a decent shot of espresso because it extracts too fast and what you’re missing there is the CO2 in the beans which give you a bit of resistance because as soon as water hits the coffee puck CO2 wants to get out and it gives you a kind of a diffused resistance over the whole puck and if you’re missing this then it will extract faster and so if you had the fine ground coffee in the filter extraction, it seemed that this degassed faster or more vigorously and this leads to a lower flow rate initially. We saw that in the later stages of the extraction the fine ground coffee caught up and even at some point overtook the medium and coarse ground, but initially like the first part of the extraction the finest was actually extracting the slowest.

Also, we’re into grinding. Up till now there’s been a lot of… When you talk about grinding what people mostly use is the average particle size, but what most of you will know is that grinding coffee actually results in two different size ranges. You always have a bit of fine and you have the course range and we’re also into kind of trying to come up with a better estimators of numbers how to characterize your grinding. First of all, we kind of measure how much fines you get. If you get 20% to fines or 30% fines as a certain grind level so we can also compare different grounds to each other which is something I think is still missing. It’s kind of hard to compare grinders against each other and also what is quite an interesting development is that Scott Rao recently had some thoughts about how to measure extraction yield and what it actually tells you because there’s two very different approaches basically. In the end we mostly do the economical approach, and this is based on measuring the TDS in your cup.

This tells you how much coffee did you get into your cup out of the of the ground coffee but it doesn’t tell you how much coffee is actually being extracted because there’s still some leftover liquids between the coffee grounds and especially for immersion extractions. So, for example French Press, you also get quite a lot of liquid retained in the coffee at the end and so basically you can choose how much, either you just concentrate on what you get in your cup. This is kind of the bang for your buck and the other version would be or the other approach would be to try to estimate how much has actually been extracted from your coffee in total. So, this would then include all the leftover water that’s still between your grounds and this will could well be a bit closer to the actual sensory impression because for example if you extract based on what you need in your cup, you extract 20% and then you go back and kind of try to push out all the remaining water between your grounds and measure these two then you could well end up at 22 or 23%. So, you actually are even though you didn’t know this when you only measured your beverage you could actually have over-extracted it and so I guess there’s going to be some more thinking we all have to do to come up with a solution on how to handle this.

I also can tell you that all data that has been shown here was gathered and interpreted by myself so there’s no other sources involved here and lastly; I want to show you a short view of where I actually work. It’s the Coffee Excellence Center headed by Chahan Yeretzian. We have five pillars here. We have Sebastian Opitz who’s also a biologist and a chemist. He has a double degree. He’s mainly concentrated on Origin so on green coffee quality. We have Samo Smrke who is mainly involved in transformation, this is mainly roasting. Myself, I’m mostly into the extraction part and sensory analysis. We have Anja Rahn, Dr. Anja Rahn, she’s an aroma specialist and we have Sabine Stauffacher who’s helping us with education. Okay. Thanks so much for attending and please don’t forget the evaluation forms. Questions anyone? Nobody. Yeah.


46:30 When plotting out TDS samples by the second, what physical methods did Dr. Wellinger use to gather this data?

Attendee 1: When you’re when you’re plotting out your TS samples by the second.

Marco Wellinger: Yeah.

Attendee 1: How are you doing that? Are you just like sticking a pipette down there and pulling one, you know every second. How do you plot? If we wanted to kind of replicate something similar.

Marco Wellinger: Yeah. Okay. We got a couple of different systems. The one I showed here was student of ours who helped with the work and basically you have… Have you ever seen this, split your espresso seven ways. You get basically a piece of wood with shot glasses in it

Attendee1: Okay

Marco Wellinger: and then they’re kind of aligned. so, the rim of one glass touches the rim of the next.

Attendee 1: Okay.

Marco Wellinger: then you’ve got the timer by the side?

Attendee 1: And then you just sliding it?

Marco Wellinger: Yeah. And then you’re just sliding it.

Attendee 1: Oh, okay.

Marco Wellinger: I mean, there’s also more sophisticated approach with kind of sample turntables. Sometimes people have used this to just put all the samples in there and then it turns automatically.

Attendee 1: Okay, cool. Thank you.


47:30 Why did Dr. Wellinger choose calcium carbonate to express acidity in coffee?

Attendee 2: I saw your titratable acidity was, you are quantifying it in terms of calcium carbonate?

Dr. Marco Wellinger: Yeah

Attendee 2: Explain why you are expressing it like that and how you relate it? So, if you’re titrating it with sodium hydroxide,

Dr. Marco Wellinger: Yeah.

Attendee 2: how are you relating that because there’s so many different acids in coffee and it’s hard to just choose one to relate it from whatever your units of sodium hydroxide to whatever you’re expressing it as? Does that make sense?

Marco Wellinger: Yeah. I think I know where you’re going with this because there’s been, I mean, there’s also the issue that people say but you have magnesium in your water. How do you express that in calcium carbonate, PPM calcium carbonate, and it’s kind of a weird unit I have to admit. The thing is that if you dissolve limestone, like calcium carbonate in water, you get two components out of it – you get total hardness, which is the calcium part, but you also get the hydrogen carbonate and the hydrogen carbonate is basically a base and then you can relate the kind of molar amount of hydrogen carbonate. You can relate too how much scale it would form, and this is what PPM calcium carbonate actually tells you. If you have 100 PPM calcium carbonate, it tells you that you get a 100 milligrams of scale per liter of water. This way you can also transform these PPM calcium carbonate into molar units for hydrogen carbonate and then you could just take it inverse. Hydrogen carbonate can, we usually use it here in coffee as a base, but you can say okay this many moles of for example, how do you call it, HCL of, what is it again? Yeah, for specific acids like sulfuric acid how many moles of the folic acid it would take to neutralize 100 PPM calcium carbonate of alkalinity. That’s how you get it in line.

Attendee 2: Oh, okay, so you’re not expressing it for a specific acid. you’re expressing it

Dr. Marco Wellinger: as total acid.

Attendee 2: as total acid

Dr. Marco Wellinger: Its total acidity yeah.

Attendee 2: That makes sense. Thank you.

Dr. Marco Wellinger: Okay. Thanks.


49:50 What type of software did Dr. Wellinger choose to express the correlation curves?

Attendee 3: Hello, just out of curiosity regarding your data collection.

Marco Wellinger: Yeah.

Attendee 3: The plan says you were collecting your data on acidity and correlating that to TDS. What type of software where you using to express those correlation curves?

Marco Wellinger: I mean the graphs itself they were all in Excel.

Attendee 3: In Excel

Marco Wellinger: Yeah.

Attendee 3: and so, you’re not using like Minitab or any type of advanced

Marco Wellinger: Statistic software

Attendee 3: statistics, yes, statistic software?

Marco Wellinger: No, actually I mean there’s, that issue is that we do so much different things that it’s usually not worth it to have something more specialized. I guess this is more suitable if you do a more routine analysis and this way we just use Excel, but I mean we do have R also for like more complex stuff we use R but in this case, it’s fairly simple.

Attendee 3: Cool. Thank you.

Attendee 4: Hey, I just wanted to thank you for your presentation. The works very interesting. My lab is actually working on TDS right now extraction and so if everyone seems pretty interested, if you want to look at the poster upstairs from UC Davis, we’re doing TDS on basket versus cone for medium and dark roast

Marco Wellinger: Okay.

Attendee 4: Yeah, it’s not espresso but it’s really interesting to see the parallel. So, thank you for that.

Marco Wellinger: Okay. Thank you too. So, you’re working on the brewing control chart?

Attendee 4: Yeah. upstairs yeah with UC Davis, Dr. Ristenpart, Dr. Guinard.

Marco Wellinger: Okay. Yeah, I’ll definitely stop by. Thanks so much. Okay, seems that’s it. Thanks so much.


50:45 Outro

Heather Ward: That was Dr. Marco Wellinger at Expo in 2018. Remember to check our show notes for a full transcript of this lecture and visit worldofcoffee.org for tickets to our next run of lectures!

This has been an episode of the SCA Podcast. Thank you for joining us!

Subscribe to the #SCApodcast on iTunes, Stitcher, Soundcloud, Pocket Casts, or RadioPublic.