Nasko Panov

# Regrinding Coffee

### Effect of coffee beans addition speed to the grinder on the particle size distribution

Dear readers, thank you once again for your support and interest in my articles. I’ve been less active for a few weeks because of personal issues. Now I am back full of ideas and ready for some coffee science.

I suppose that everyone has watched the video from James Hoffmann about Regrinding coffee. If not, you can follow the link __HERE__. It is about the difference in coffee grind size distribution which appears as a result of the speed you use to add single-dose coffee to the grinder or grinder hopper. This is an interesting phenomenon that has some theoretical explanation. Namely that the difference is due to the time period the coffee beans spend in the grinder while grinding. Adding the beans slowly will decrease the time that a single bean spends in the grinder resulting in a hypothetical coarser grind. This could influence the coffee grinds uniformity and grinding consistency. And this is exactly what is observed in practice. Adding the beans one by one generates a particle size distribution (shortly PSD) that is increasing the flow through the coffee puck resulting in quicker extraction time. In order to achieve the same brew time (for espresso), you must grind the coffee much finer.

* Please note, that this experiment was only about grinding whole beans and not about grinding coarser and then regrinding finer. I chose the name “Regrinding coffee” because I wanted to link that we are talking about the same topic with James Hoffmann, and my article is a kind of scientific explanation or extension to his video.

I decided to take a deeper look at what is happening when adding coffee slowly and what is the difference in the PSD compared with the standard grinding techniques.

And what is more appropriate to examine the coffee particle size than using a Lased Diffraction Particle Size Analyzer?

In parallel with the main goal – to see the difference in the PSD of coffee added slowly to the grinder compared with coffee added at once, I wanted to observe if there are additional variables that could influence this distribution. I tested four different grinders – one with conical burrs and three with flat burrs, but also tested different roasting profiles – light, medium, and dark roast. Here I’d like to give my appreciation to one of my Patrons - __@blazarovbg__ for providing me with additional grinders and helping me collecting all the samples.

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. Once the analysis is complete the user has a variety of approaches for reporting the results in a numerical or graphical view. Some might prefer a single number answer – what is the average size?

But in practice, a single number can’t describe the PSD of a sample.

The graphical representation is also often difficult to interpret as the differences are sometimes small to visualize but meaningful in practice. Adding the possibility that one result could be represented on a linear or logarithmic scale becomes even harder.

First, I’ll start with the numbers. I said that there is no simple numerical answer to the question – What is the PSD? Or, at least the full understanding of the result can’t be represented as a single number. The answer is much more complex and analyzing every part of the result could give us information about the process happening in the grinder and later during coffee extraction.

Distribution Width is one parameter that could provide us with information about how good the grinder is doing its job. The best grinders should create a narrow distribution with the smallest width. This means that the coffee particles are more uniform resulting in uniform coffee extraction. One of the common values of width used for Laser diffraction result is the Span shown in the equation below:

An additional 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). Here are the Graphs with all results for Span and COV of all measured samples:

Visible from both graphics for COV and Span is, that slowly adding coffee beans to the grinder generates smaller distribution width (orange bars) compared to adding the beans all at once (blue bars). It is valid for all grinders used, both types of burrs (flat and conical), and for all types of roasts. Narrower distribution of the particles generates better and more unified coffee extraction. Narrowing the distribution means reducing the number of coffee particles creating over-extraction, under-extraction, or uneven extraction in general.

This experiment and results are a prove of something that was already observedin practice – slow addition of the coffee to the grinder leads to higher extraction yields, better crema, and overall taste.

Distribution width, however, couldn’t answer the question – Why slowly adding coffee to the grinder leads to faster espresso shot/higher flow rate?

To answer this question, we need to look at another parameter – the Median value. The median value is defined as the value where half of the population resides above this point, and half resides below this point. It is also called D50. The D50 is the particle size in microns that splits the distribution with half above and half below this diameter. This value is one of the most meaningful for PSD. Let’s see what we get from the experiment:

When coffee beans are added slowly to the grinder results show a higher D50 value (orange bars) with only one exception – the Eureka Mignon 50mm Flat Burr. I think this is more likely to be a measurement error or sample preparation mistake than a grinder issue.

Results are more than clear – a higher D50 value meansacoarser grind.

This could explain the observed fact that in order to achieve the same espresso shot time, the grinder must be adjusted to a finer grind setting.

There is another thing I found interesting in this experiment. It is the D10 value – the portion of particles with a diameter smaller than this value. It is the portion of coffee particles with a size lower than 50-100 microns called fine particles. Finer particles in the coffee mean a bigger chance of basket clogging and reducing the flow. Fine particles are also able to pass through the basket ending in the final espresso drink, making it harsh and probably unpleasant. Fines are also related to over-extraction as they are most easily extracted.

The results from the Laser diffraction are definite. There is a clear difference in D10 value for grinder loaded with coffee at once compared with slowly adding the coffee beans to the grinder/hopper. The effect is valid for all grinders, burrs, and roasts. Clearly, a better result for grinding is achieved by adding coffee beans slowly to the grinder.

Doing it that way is increasing the D10 value meaning more uniform PSD with fewer fines or fines with bigger diameter.

However, how slowly? James Hoffmann proposed that the beans should be added one by one making the grinding process horribly slow and annoying. But according to the measurements I made, I can’t support this claim. I experimented with adding 18g of coffee for different time periods – 30s, 40s, and 50s. I found that in the above 30s all samples behave almost the same. Unfortunately, when I was planning this experiment, I didn’t expect that this parameter will be so important, and I analyzed a limited number of samples. So, I’ll leave this analysis for the future, when I’ll prepare more samples in shorter time intervals and will analyze the results in-depth. Then I’ll answer the question – How slowly? Looking at the initial results, it seems that there is no need to add the beans one by one. Adding them little by little (5-6 beans per batch) and providing them enough time to exit the grinder is enough. I clearly understand that the 30s mark I mentioned is a value valid only for the grinder I used – Baratza Sette 270 and other grinders may perform differently.

Here are the linear and logarithmic graphics representations for a few samples. As I said, it is hard to analyze the PSD based only on the picture, but after understanding the numbers standing behind these pictures, it is much easier:

In all cases we are observing the same thing – better grinding results are achieved through slowly adding coffee beans to the grinder. This difference in the performance is measurable and it has a significant effect on the espresso preparation and extraction result. According to this experiment, it is always preferable to add coffee slowly for the best results.

As a summary, it generates a narrower PSD width (Span and COV), Higher D50 (meaning coarser grind), and fewer fine particles (D10).

It is always great when there is a clear experimental explanation of the observed effect. I’ll continue working on this topic in the future answering more questions:

What is the appropriate beans addition rate and where is the limit when the results get worse?

Is this effect so clear on high-end coffee grinders like ЕК43?

Are there other factors influencing PSD?

Until then, stay safe and prepare your great coffee slowly.

*If you like the article you can support on my Patreon page *__HERE__* or visit *__www.patreon.com/npcoffeescience.__* 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!*

*If you like the article you can support on my Patreon page*

__HERE__*or visit*

__www.patreon.com/npcoffeescience.__*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!*