Different effects influencing Particle Size Distribution in manual grinders
This is a continuation of a project I started almost a year ago. In the next few months, I’ll publish a series of articles related to coffee grinding and the main idea is to better understand the physics and effects of grinding using several of the widely used manual grinders on the market. This won’t be a product review or product comparison, but every grinder will be put to a test and the results will show how they are performing. Keep in mind that the main goal is to find dependencies, effects, and correlations that are influencing the consistency and performance of the grinders. To have better statistical results several grinders will be used in this series of experiments.
The first thing I wanted to test was to transfer my previous experiment to a manual grinder. If you haven’t read my previous experiment, please follow the link HERE to learn more on the topic. It showed some remarkable results for grinders and the effect of coffee bean addition speed. It was assessed that the slow addition of the coffee to the grinder is leading to narrower distribution of the particles leading to better and more unified coffee extraction. The experiment clearly shows that better results are achieved using this approach. This difference in performance is measurable and has a significant effect on the espresso preparation and extraction results. According to this experiment, it is always preferable to add coffee slowly for the best results. As a summary, it generates a narrower Particle Size Distribution width (Span and COV), Higher D50 (meaning coarser grind), and fewer fine particles (D10).
The previous experiment was conducted with Laser Diffraction – a powerful tool for analyzing the Particle Size Distribution (PSD) of a sample. The same instrument will be used for all future articles related to the performance of the manual grinders I’m testing. All the data will be processed because in Particle Size Analysis the most challenging part is the evaluation of the results or extracting valuable information from all the numbers that the instrument provides. Understanding PSD is a very challenging task. PSD is not just a single answer, a number, or a graph. The answer is much more complex and analyzing every part of the result could give us information about the processes happening in the grinder and later during coffee extraction.
Flair Espresso was the first company to respond to my project. After several e-mails, they send Flair Royal – a manual grinder from their portfolio. And this will be the first grinder I’ll put on the test. Two of the most important questions and the ones I’m truly interested in are:
1. Does adding the coffee little by little to the grinder influence the PSD, grinding performance, and consistency?
2. Does the speed of rotation influence the PSD, grinding performance, and consistency?
Together with these two important experiments, I’ll test how good the grinder is performing in general, but more thorough testing will be left for future experiments and articles.
Coffee addition speed has a huge impact on PSD and grinder performance when working with an electrical grinder as previously outlined. Prior to this experiment, I was thinking that coffee addition speed could have little to no effect on manual grinders because the rotation speed is very low compared to electrical. Grinding speed for an electrical grinder could vary from 350 RPM up to 1500 RPM. Using a manual grinder with a lot of effort I can achieve around 120 RPM, or 3-10 times slower. This was a premise for a lack of an effect observed in the electrical grinder caused by the coffee bean addition speed. The main explanation for this effect in an electrical grinder is the pressure that the beans entering the grinder are exerting on the particles already inside (the “pressure effect”). This could lead to additional beans cracking and fragmenting. Another proposition was made for this effect – the time the upper beans spend in the grinder being a grind. If there are many smaller particles already inside, this could cause something like a “clogging effect”, slowing the speed the coffee is passing through the grinder. This will extend the time for a single bean spending in the grinder, leading to bigger chances for the bean to be crushed to small pieces.
Both hypotheses require high grinding speed – much higher than a regular speed used in manual grinders. Nevertheless, I prepared an experiment to test it. I used Flair Royal manual grinder set in three different grind settings – fine, medium, and coarse. I added 10g of coffee. After that I cleaned the grinder and using the same grind settings, I started adding portions of 3-4 beans of coffee, grinding it completely and then adding another portion of 3-4 beans.
Before continuing with the results, I should explain first what D10, D50, and D90 mean. D10 is a particle size in microns below which 10% of the particles have a smaller size. D50 is a particle size in microns that splits the distribution with half above and half below this diameter. D90 is a particle size in microns below which 90% of the particles have smaller size. Here is a graph (Pic 1) to represent the values:
Pic 1: A representation for D10, D50, and D90 values in a particle size distribution.
D10 is the portion of coffee particles with a size lower than 50-100 microns called fine particles. The consistency of a grinder is also related to the D10 value as this parameter is showing how many fine particles the grinder is creating. D50, also known as a Median value. It is showing roughly how fine or how coarse is the overall grinding.
Now let’s compare these values for Flair Royal manual grinder obtained in standard conditions compared to when the beans are added little by little. Below are the results for D10, D50, and D90 related to fine grind settings:
Pic 2: D10, D50, and D90 results for Flair Royal manual grinder at fine grind settings.
As seen in the above chart (Pic 2) there is a significant difference in all 3 values for D10, D50, and D90. The blue graph is showing the values when 10g of coffee is added at once to the grinder. The orange bar is showing the result when the coffee is added on little portions of 3-4 beans to the grinder. Comparing the results, the second method creates an overall coarser grind distribution in the whole range. If you compare these results with my previous experiment for an electrical grinder, contrary to the expectations this effect is even stronger. In the D10 range, the difference could reach 35 microns compared with an average of 25 microns difference in electrical. Of course, this is an observation only for that manual grinder, but coming to my future articles I’ll update this experiment with results from other grinders. The next picture is showing the PSD in a graphical way:
Pic 3: Comparison of PSDs for Flair Royal at fine grind settings.
Little disturbances could be observed for the orange graph representing the PSD achieved with the coffee beans slow addition technique. Of course, this result will be updated when subsequent articles and results are going to be published. One possible explanation for the observed fluctuation of the result is the amount of sample used for the laser diffraction analysis. If the coffee sample is too small, sometimes the number of particles is not enough to “smooth” the graph. Even though, the result can be used for data analysis as the experiment gave good repeatability and consistency.
Here are the results for the same experiment with Flair Royal on medium and coarse grind settings showed on Pic 4, and Pic 5:
Pic 4: D10, D50, and D90 results for Flair Royal manual grinder at medium grind settings.
Pic 5: D10, D50, and D90 results for Flair Royal manual grinder at coarse grind settings.
The overall coarser grind achieved from the grinder when the beans are added little by little is also visible from the so-called Mean, Median, and Mode results. These are the basic terms describing some of the key characteristics of the particle size distribution curve. Mode is the value where the peak diameter from the curve is situated or the most commonly occurring particle size. Median is a particle size in microns that splits the distribution with half above and half below this diameter, also known as D50. Mean is the average value for the distribution. Here is a graphic showing all the values on Pic 6:
Pic 6: A representation for Mean, Mode, and Median values in a PSD.
The Median and the Mode values have some limitations when more than one particle population is present in the sample. Several distributions could have the same values for Median and Mode but could have totally different distribution curves. That is why these two are limited to represent the PSD, especially when they are expressed with a single number or figure. Nevertheless, the Median value is often used to compare two or more samples with a single number value representing the distribution. Keep in mind not to confuse the Median with the average particle size. This is true only for a perfectly symmetrical distribution. Here are the values for Mean, Mode, and Median for the same experiment at all three grinder settings – fine, medium, and coarse on Pic 7, Pic 8, and pic 9 respectively.
Pic 7: Mean, Median, and Mode results for Flair Royal manual grinder at fine grind settings.
Pic 8: Mean, Median, and Mode results for Flair Royal manual grinder at medium grind settings.
Pic 8: Mean, Median, and Mode results for Flair Royal manual grinder at coarse grind settings.
Slowly adding the coffee to the grinder is creating a PSD with higher values for Mean, Median, and Mode, meaning an overall coarser grind. Contrary to the proposition made earlier, this effect of achieving more coarse distribution was observed in a manual grinder. There should be a third explanation together with the “clogging effect”, and “pressure effect”. I have some ideas about that but will keep the explanation for another article.
As stated in my previous article HERE, Distribution Width is one parameter that could provide us with more information about how good the grinder is performing. The best grinders should create a narrower distribution with the smallest width. This means that the coffee particles are more uniform resulting in more even coffee extraction. Some may argue about that and claim that at a certain point if the distribution is too narrow the extraction and coffee taste will start to degrade. This could happen in theory, but in reality, a manual grinder can’t create such a narrower distribution. This proposition could be true only for the best performing electrical grinders with the narrowest distribution width, and still, this is a big debate among coffee experts. In the common case, the grinder with the smaller particle PSD width will perform better and will produce better-tasting coffee.
One of the most used values of width used in the Laser diffraction experiment is the Span. This is an indicator of the distribution width. It represents the distance between two points equally spaced from the median value. Here is the formula for this parameter on Pic 9:
Pic 9: Formula for Particle Size Distribution width called Span.
An alternative approach to describe the distribution width is to normalize the standard deviation through division by the mean value. It is called the Coefficient of Variation (COV). It shows the spread of the data, but it is normalized, meaning it can be used to compare distributions with different Mean values. In my experiments, this parameter is one of the most indicative because we already saw that adding the coffee beans on portions to the grinder leads to PSD with a higher Mean value. My previous experiment with electrical grinders showed a significant difference in Span and COV values. It concluded that adding coffee slowly to the grinder generates smaller values for Span and COV, hence the distribution width is narrower and the grinder performance - better. But for manual grinders, this is not the case and the result could be seen on Pic 10, Pic 11, and Pic 12:
Pic 10: COV and Span result for Flair Royal manual grinder at medium grind settings.
Pic 11: COV and Span result for Flair Royal manual grinder at coarse grind settings.
Even though the Mean value is higher when the coffee is added little by little to the grinder, there is no major difference in the COV value for all three grinder settings – fine, medium, and coarse. This means that the effect of the slow coffee addition using the manual grinders is different than the one using electrical grinders. Still, slow addition is creating overall coarser grind size distribution and slightly narrower PSD width. It affects the grinder performance but to a much lesser extent. There is a small difference in the medium and coarse grind setting, but still not as significant as the one in the electrical grinders’ experiment. This must be related to the reduced rotation speed used in manual grinders as mentioned above.
Comparing the results from the medium and coarse grind settings reveals the COV values are similar with a difference of about 5%. On the opposite side is the result for fine grind setting - practically there is no difference in the COV values for the fine grind settings:
Pic 12: COV and Span result for Flair Royal manual grinder at fine grind settings.
Based on the achieved results, three conclusions could be extracted:
Adding the beans slowly to the grinder is creating an overall coarser grind distribution visible from the values for D10, D50, D90, Mean, Median, and Mode.
Adding the beans slowly to the grinder is leading to PSD with a slightly disturbed curve, but this result must be thoroughly examined as this could be the problem with the performed test and sample amount used.
Adding the beans slowly to the grinder is affecting the PSD width to a much lesser extent, hence not affecting the overall grinder performance. The difference in COV values is similar with slightly better PSD width achieved on medium and coarse grind settings.
An additional conclusion could be made about the Flair Royal manual grinder – based on the COV value, the grinder is performing better than the tested electrical grinders from my previous experiment. It showed stable performance and consistency in the entire grind settings range including the espresso range. In the future I’ll test few more things with this grinder, so stay tuned for future articles.
While researching and writing this article I found few more parameters that could be calculated to evaluate the effect of coffee addition speed to the grinder - the Interquartile coefficient of skewness (Skew), and Kurtosis. These, together with the conclusion about the second question I asked at the beginning of this article will be discussed in a second article soon.
Disclaimer: I’m not financially motivated by the company that provided the manual grinder. Flair Espresso offered me the Flair Royal grinder without obligations or any expectations. All the experiments discussed results, and conclusions are my own and are not financially motivated.
You can still support me on my Patreon page HERE, visit www.patreon.com/npcoffeescience or click the red button below. At the price of a single espresso per month, you can read all my experiments, work-in-progress projects, future ideas, and allow me to continue my work. Thank you!