Flour Milling Lab Results
In order to fine tune my milling and sifting process, I ran a series of tests at different mill coarseness settings to see which setting might result in the best separation of bran from endosperm. I then ran a successive reduction multi-pass milling and sifting process at what appeared to be the best first pass settings and sent samples of all these tests to CII Labs to see what some of the ash content, protein content, and dough rheology might be. I also sent in some samples of Heartland Mill flours to use as a reference, since these are the types of flour I would like to emulate with my home milling process. In fact, I use Heartland Mill Hard Red Spring Wheat Berries in all these milling tests.
The equipment and general process has been described in previous blog entries on home milling and sifting.
The processing is accomplished using a Meadows 8 Inch Stone Mill and a Meadows Eccentric Sifter, as well as a sieve shaker that can stack several 12 inch diameter stainless steel or brass sieves of US standard sizes.
The CII Lab results for the initial test of varying coarseness settings of the mill have sample descriptions such as "P1 open 1/3 turn 35-50". The P1 is just a label for the setting of the mill, which is listed next. At 1/3 turn, the mill stones are separated by about 1/2 a grain width. Similarly 1/6 turn would be about 1/4 of a grain width, and so on. The numbers at the end refer to the size of the sieves used. So, a sample labeled 35-50 went through a US standard number 35 sieve and was caught in the US standard number 50 sieve.
The CII Lab results for the second test start with a first pass with a 1/6 turn opening, which seemed to be a good setting to get good initial separation of bran from endosperm in the initial tests to determine the best first pass mill coarseness setting. The sample descriptions have labels like "P2a <70" or "P1 40-80m". The P2a is the label of the pass in the process described in the process flow chart for this milling session. The "<70" refers to product that went through the US standard number 70 sieve in the stack of sieve shakers used for all but the first pass.
On the first pass, "P1", the Meadows Eccentric Sifter was used. It has sifter sections specified by the mesh size of the screens in the three sections of the sifter. So, the "40-80m" refers to flour that went through the 40 mesh section and was caught in the 80 mesh section.It so happens that due to the wire diameters of the material used in the screens, the 26m section has about the same opening size as a US standard number 20 sieve, the 40m section has about the same opening size as a US standard number 35 sieve, the 60 mesh (not used in this session) is about the same opening size as a US standard number 50 sieve, and finally the 80 mesh screen has the same opening size as a US standard number 70 sieve. So, in order to simulate the use of the Meadows sifter in subsequent passes with my smaller sieve shaker that is just more practical for these smaller amounts, I used US standard numbers 20,35,50, and 70 sieves in the stack.
I also include a spreadsheet (xls, html formats) that summarizes the results of the milling session. On the "model" sheet, there is an attempt to model what would happen at other mill settings than the one I used, based on data from the initial runs at various coarseness settings and my guesses about how the subsequent millings would go. All of that may not be very useful, except to me. However, the "model" sheet also shows a summary of the basic ash content output of each stream from the process I ran.
The flour streams with ash content around 1% had very reasonable rheological properties, which makes sense, since I was able to make some very nice breads very similar to what would have been possible with Heartland Mill Golden Buffalo flour. The stream of lower ash content flour seemed to have low mixing tolerance, so I must have inadvertently separated out some important components of proteins needed to form good quality gluten. This tells me I can create a flour with an ash content of somewhere around .85% by mixing some of the other higher protein streams with the very low ash stream to get an off-white flour that is whiter than Golden Buffalo, yet still will create a strong enough dough. I suspect that using the lowest ash stream by itself might result in a dough that doesn't have the best baking properties, since the farinograph showed weak mixing tolerance relative to the other Heartland Mill products or my own higher ash content flours more similar to Golden Buffalo.
Additional Results From Wheat Montana Berries Milling Session (added 2/26/08)
I conducted a similar series of milling tests with Wheat Montana Bronze Chief berries, which are hard red spring wheat berries. I wanted to see if the process would go differently with the harder berries and if this would suggest changes to the process of the mill settings and sifting approach.
A flow chart of this milling session, very similar to the last, other than the addition of a fourth pass, has been posted. The preliminary report from CII Lab (now updated to final as of 3/6/08) is also posted. The nomenclature for the various passes is similar to above. However, the various tests of the "first pass" are labeled with letters. For example, a label of "PA 35-50 1/12 Turn" refers to a first pass test using a mill setting of 1/12 turn (1/3 turn is about 1/2 berry width in the separation of the stones, so 1/12 Turn would be a stone separation of about 1/8 of a berry width), and the 35-50 refers to product that fell through the #35 sieve and was caught in the #50 sieve.
The numbered passes refer to the multipass milling process in the flow chart above, which was used to create various grades of flour, bran, and red granular product.
I also include an updated version of the spreadsheet (xls, html formats) mentioned above that summarizes the results of the Wheat Montana milling session in two addition sheets ("WMactual" and "WMmodel"). On the "WMmodel" sheet, there is an attempt to model what would happen at other mill settings than the one I used, based on data from the initial runs at various coarseness settings and my guesses about how the subsequent millings would go. All of that may not be very useful, except to me. However, the "WMmodel" sheet also shows a summary of the basic ash content output of each stream from the process I ran. The "WMactual" sheet summarizes the actual results for the settings used, but it adjusts for the fact I removed some of the intermediate products to send in as samples for testing at the lab. The model is not exactly what I did, but it is a better description of what would happen if the intermediate samples had not been removed as I conducted the milling session.
Overall, what I discovered is that a finer setting on the mill seemed neccessary to get about the same breakout between "bran", "coarse red granules", "coarse white granules", and "cream flour", which is a rough way of describing 4 products that seem to be produced when I mill berries using a fairly coarse setting for the mill (1/6 turn of the screw from when the stones just touch, corresponding to about 1/4 of a grain width).
Also, I am becoming aware of the fact that there is a threshold effect in the mill setting that has a big impact on the separation of the bran and the yield of white granules that seem to then yield flour with the lowest ash content when remilled and sifted. Only the slightest change in the coarseness setting around the 1/6 turn setting for HRW or at about the 1/8 turn setting for HRS berries seems to make an enormous difference in the relative yield of "white granules" in the first pass of the mill. If the initial pass is too coarse, then result is much more bran attached to large granules, and if the setting is too fine, then the result is too much high ash content flour in the first pass, and less yield of white granules, consequently reducing any chance to extract lower ash content flour in the second milling steps.
I was also struck by the variation in protein and ash content of the berries sent in for testing. I'm wondering if there is a more accurate test or larger sample size needed for the grain in order to get more consistent results. The ash content and protein levels for the berries weren't as expected. For example, the HRS berries (Bronze Chief) from Wheat Montana should have had a higher protein content than the results on the tests show. Also, the HWS berries (Prairie Gold) had an inordinately low ash and protein content than what is normally said to prevail with this type of wheat berry. So, either the tests aren't revealing the true levels for some reason, or the wheat berries vary much more than I thought. Unfortunately, this will require convincing someone at the lab or another expert in this somewhat esoteric area to take a charitable interest in educating me.
Tempering may need to be adjusted for these berries, and I may have learned something new about the tempering time. First of all, it seemed to me that the berries milled as if they needed a little more moisture content. The amount of ash is larger overall. My sense was that the berries and the flour seemed dry. Although I added enough moisture to reach a moisture content over 14%, the moisture content tested at only 13.3%, which may have resulted from letting the berries sit for a little over 48 hours. The tempering period may have been long enough to allow some moisture to escape. Possibly these harder berries have trouble absorbing the moisture, which makes it more available to evaporate from the surface over a period of time. The lids on my tempering containers are probably not perfectly air tight, so moisture may escape very slowly over a period of time. Next time, the tempering for Wheat MT HRS and HWS berries will be conducted in two steps. First, enough moisture will be added to bring the berries to 14% moisture content. Then, in a subsequent step about 24 hours later, the berries will be tested and enough moisture will be added to bring the berries to around 14.5% moisture content followed by an additional 24 hours before milling the berries.