Ed Wolfrum, Feedstock-Conversion Interface Consortium Principal Investigator: I think we’re at the top of the hour. Erik, did you want to start the recording?
Erik Ringle, National Renewable Energy Laboratory: I did. It’s going.
Ed: Ok. Excellent. Ok. Well, thank you everyone for joining us today. Welcome. This is the fourth in a series of webinars about the Feedstock-Conversion Interface Consortium, the FCIC. My name is Ed Wolfrum, and I’m fortunate to serve as a principal investigator of the FCIC. I’m glad you’ve decided to join us for an hour today. Now, the FCIC is led by the Department of Energy’s Bioenergy Technologies Office, BETO. And that’s a collaborative effort among researchers from nine different national laboratories to address the challenges posed by feedstock and process variability and bio refineries.
Like I said, this will be the fourth in a series of webinars. In December, I presented an overview of the FCIC. In February, FCIC researchers shared some work on computational modeling approaches to address feedstock handling issues. We took a break in March, and then in April, other FCIC researchers discussed issues that are associated with the intrinsic variability of feedstocks. Now, a recording to all those webinars are available on the FCIC website, and today we’ll have a link to that, and today’s talk will also be on the webinar in a couple of weeks. Note that these webinars are going to go on hiatus after this one for the summer. We’ll start up again in the fall some time in September.
For those of you who missed the first webinar, I’m going to spend about a minute talking about the FCIC in about a minute. But the main event today are presentations by two FCIC researchers: Dr. Jun Qu of Oak Ridge National Laboratory and Dr. George Fenske of Argonne National Laboratory, who will talking about understanding and mitigating the issues associated with wear in biomass combination systems. Now, because of the uncertainties in virtual presentations that we’ve all experienced for the last year, we’ve recorded that talk. So there will be a little bit of a break after I finish my talk until I get that one going.
A couple housekeeping issues. Everybody is muted. Any problems with sound or video please put in the chat. My colleague Erik Ringle will be handling those. And again we are going to make a recording of this presentation and it will be available. We have to caption it and get it online, takes a couple of weeks. Now I do have to read a disclaimer so please bear with me for a minute. This webinar, including all audio and images of participants and presentation materials, may be recorded, saved, edited, distributed, used internally, posted on DOE’s website, or otherwise made publicly available. If you continue to access this webinar and provide such audio or image content, you consent to such use by or on behalf of the DOE and the government for government purposes and acknowledge that you will not inspect or approve, or be compensated for, such use.
So with that, let’s move to what is the one slide guide to the FCIC. As I said, we are a consortium led by BETO, a collaborative effort among researchers from nine different national laboratories. We focus on two key ideas: biomass feedstock properties are variable and different from other commodities, and empirical approaches to address these issues haven’t been successful. So biomass feedstock properties are variable. They change over time and space depending how material is harvested, transported, stored, and of course processed. And that variability can be between species like corn stover or switchgrass, or within the same species. And it’s affected by all of those variables.
When I say properties are different, the properties of biomass like the corn bale or the corn stover bale you see in this picture to the right, or the wood chips in the bottom right corner, those are different from the more familiar agricultural commodities like corn, wheat, rice, soybeans that have had about 100 years’ head start on biorefineries for handling and storing and processing. The empirical approaches that have been taken to date to try to adapt some of those learnings from those agricultural commodities to biomass feedstocks haven’t been entirely successful. So the FCIC is taking a first-principles approach. We’re developing first-principles-based knowledge and tools to understand and to mitigate the effects of biomass feedstock and process variability on biorefineries. You can—there’s a website down at the bottom of the corner. You can come and see all the details about what we are working on.
Today’s speakers, like I said, the main event is a talk on the impact of wear on biorefineries. Dr. Jun Qu and Dr. George Fenske are kind of doing this as a team. Dr. Jun Qu is currently a distinguished R&D staff scientist in the materials science and technology division at Oak Ridge National Laboratory. He’s published three book chapters and over 110 peer-reviewed journal articles, has 13 patents, and has produced two ASTM standards and been recognized by a handful of external awards. Dr. George Fenske is the tribology lead for Argonne National Laboratory’s applied materials division. He has more than 40 years of experience as a PI on research projects in tribology of hard coatings and surface modifications and is an author of over 100 different publications.
I’ll—a little bit of a plug. Jun and George bring a wealth of knowledge not typically associated with our refineries into the FCIC to address a very specific industry problem, and I think that's one of the strengths. Together they lead the materials of construction task to work to understand how to measure, predict, and mitigate wear in processing equipment that are relevant to the emerging biorefinery industry. So during the presentation, like I said, please use the chat box to ask questions. My colleague Amy Slider, who is the FCIC project manager, will be watching those. At the end of the presentations, Jun and George, who are on the call now, will be available to answer these questions. So thanks for your attention and let me—give me a minute here while I queue up the other presentation.
Erik: Hi Ed. It looks like you might need to share your computer audio in order for us to receive the audio as well.
Ed: I did that again.
Erik: That sounds great.
Jun Qu, Oak Ridge National Laboratory: Hello everyone. Today we’re presenting on recent technical progress using combined approach of characterization, modeling, and material development to understand and mitigate the issues in commercial and state-of-the-art biomass size reduction systems. This slide shows some example commercial milling machines to the left, including hammer mill, knife mill, disc mill, and ball mill, and some of them are being used widely in biomass feedstock size reduction. And to the right shows a Crumbler rotary shear, which was recently developed by Forest Concepts with the aid from the BETO program. In this study, we’re trying to use characterization modeling and the material development technologies to understand the fundamentals that wear on feeder mechanisms and develop cost-effective, resistant tool materials and also optimized tool designs. And as a result, we want to improve the tool life, increase the throughput, and also reduce the downtime and power consumption.
Before we get into the specific wear model analysis for the hammer mill and knife mill, I want to give you a quick intro of the general type of wear. There are a lot of different types of wears from the mechanical side and also the chemical side. But for the size reduction equipment, the most two important ones are erosive wear and abrasive wear. The erosive wear—it’s produced by bombarding hard particles onto the tool surface. And it’s often recognized as process deformation and microfracture. So therefore, the most important mechanical properties of the tool material is fracture toughness and hardness.
In contrast, the abrasive wear is produced by two moving bodies. It then, for two body abrasion, either the second body with the hard surface asperities or some trapped hard particles without motion scratched the tool surface. In three-body abrasion, the two bodies are not in direct contact, but there are some hard particles trapped in between at the interface. They are rolling or sliding to remove the material of the tool surface. So therefore, often we can observe grooves, cutting chips, and grid fracture on the tool’s surface. So as a result, the most important attributes of the tool material would be hardness and yield strength.
This work started with hammer mills about a couple years ago. And because using blunt blades, the major size reduction mechanism is crushing. So therefore, erosion has been identified as the dominant wear mode for the hammer mills. So there are a couple of hammer mills in ORNL we’re starting to study. The stage one hammer mill used the bear low-carbon steel hammers. Stage two hammer mill actually used the thick wire overlay of tungsten carbide iron composite to protect the surface of hammers. However, both hammer mills experienced excessive wear in processing dirty feedstocks. Then the contact pressure analysis identified it’s really the larger-size mineral particles trapped on the biomass feedstock was a major cause of the wear of the hammers.
It’s well accepted that the inorganics characterized by biofeedstock play a critical role in the tool wear. So therefore it’s important to characterize their species and particle size. However traditional method in the industry is to burn the biomass in the furnace, then collect the ash as—using the ash content as the indicator of how abrasive the feedstock is. This method, however, changed the biomass species because of oxidization and thermal decomposition. Therefore, this study ORNL developed a composition preserving extraction method. In this method, we use vibration saving and also water sanitation to separate the extrinsic minerals from the biomass. Then we use the microscope to characterize the particle size distribution, as was shown as for example to the left corner chart, which particle size from a few microns to a couple millimeters. Then we use the diffraction to determine the composition or the species of minerals. In this case, the majority is silica. Then there are some minor minerals as well.
In addition to extrinsic minerals, we also characterize the intrinsic organics of selected feedstocks. Here shows the results for the pine and anatomical fractions. In the first, by using diffraction, we learned that the needle and bark contains combined crystalline and amorphous silica, which explains why they are abrasive. In contrast, the twig, cambium and whitewood only contains amorphous, no crystalline silicon compounds. Then in addition to the species we identified for the intrinsic inorganics, we also were able to use cross-sectional SEM and EDS chemical analysis to map where exactly those second compounds locate inside the tissues, as seen to the right of the pictures.
And ORNL conducted a test by blasting pine residue onto the coupons made of actual hammer steel and results are plotted to the left. So evidently, the general trend is the higher ash content of the biomass sample the more of the tool wear that was produced. However, the trend is not linear or proportional. For example, the needle has highest ash content. However, the wear it produced much less compared to bark. So why is that? Let’s look at the chart to the right. We compare the total ash content versus how much extrinsic minerals we collect from each anatomical fraction. And evidently the needle has the lowest ratio of the minerals versus the total ash. And in contrast, the bark has the highest amount of extrinsic minerals. That explains why the bark is more abrasive.
Then if you look at the bark, cambium, and whitewood, those three have similar ratio of the extrinsic minerals versus total ash. And they have very linear correlation from the modeled ash versus the total wear. So and on the other hand, we can compare and say needle versus cambium or whitewood, they have similar amount of extrinsic wear, extrinsic minerals. However, the wear of the needle evidently produced more wear because the higher total ash. So that means we have two components here. One is both intrinsic and extrinsic inorganics are abrasive. They are causing wear. No doubt. But secondly, the extrinsic minerals are much more abrasive than the intrinsic inorganics.
Now by understanding the ash content, or especially the extrinsic inorganics were mainly responsible for the tool wear, we were looking into mitigations. And the first approach was biomass modification. And ORNL conducted a test to evaluate three methods including air clarification, size separation, and water washing. And the results are summarized in charts shown here. And clearly, the air clarification and the size separation worked pretty well to reduce the ash content and also reduce the wear. So the wear reduction was about 50% for the pine residue and 75% for the corn stover. And the cost was rather reasonable, less than $1 per ton. In contrast, the water washing method didn’t work out because evidently the water only moved the extrinsic minerals around but didn’t truly remove them.
Because hammer mill uses blunt blades, and its main combination mode is crushing. So therefore when processing biomass, a lot of fines are basically produced, which are non-useful in commercial. So therefore, by discussing with BETO and FCIC, the team had decided shifted effort from hammer mill to knife mill in FY20 Q4. Knife mills use sharp blades, so therefore the combination modes are combined cutting and crushing. So the fines are much less. Therefore more simple for biomass processing. As a result, the wear mode is different. The hammer mill is primarily dominated by the erosive wear, but the knife wear has combined abrasive wear and erosive wear.
In order to improve the knife blade performance and durability, we are looking to candidate resistant coatings and separate treatments. There are multiple criteria for that. For example, we need high hardness to reduce the abrasive wear and we need high fracture toughness and fatigue ductility to reduce erosive wear. To achieve all three attributes actually is quite challenging. Based on what were decades experience, we have identified three candidate coatings. And the first one nickel boriding, which is a thick coating up to 100 micron, the hardness about 30% or 40% higher compared to the current tool steel used by the knife blades.
Then the second was the iron boriding, which is a case hardening process. It can precise the surface layer from 100 to 300 micron deep and with hardness as high as doubled compared to the current tool steel. Then we're also looking to a thick diamond-like carbon coating, which provides much higher hardness—as high as four times higher—than the coating thickness, can be from a few microns all the way up to 100 microns thick.
We are currently evaluating the resistance in both abrasion and the erosion test because for knife mill, both the resistance to the abrasive wear and the erosive wear are important. The tool body test was conducted using ASTM standard at ORNL. And the erosion test was conducted at INL using a blasting test. The results are summarized in the two charts below. Clearly you can see the iron boriding improved the resistance for both the erosion and abrasion. However, the nickel boriding didn't work really well. So therefore, we identified iron boriding is a good candidate. And also, we after talking with the vendor of the iron boriding, so we did a calculation about cost. So that's only about 7% compared to, increase compared to the baseline. So therefore, the iron boriding can potentially provide significant tool life improvement at a really small add-on cost. And we just received the thick [inaudible] coating, which is another candidate, and results will be reported later.
After candidate materials and coatings being identified by bench-scale testing, next step is to fabricate prototype knife blades and to conduct knife mill testing. Right now INL is working with Eberbach to set up a small knife mill at INL and also acquire commercial knife blades of the materials. And Argonne and Oak Ridge are working with coating vendors to fabricate prototype knife blades. And also INL is planning the knife mill test to validate the improved performance and the tool life for those prototype blades.
After talking about the wear issue and mitigation for the hammer mill and knife mill, now let's shift the gear to the Crumbler rotary shear, the work jointly with Forest Concepts. And this new system has a lot of benefits compared with the conventional size reduction equipment. It produced more precisely controlled particle size. It reduces a lower amount of fines. And it reduced the particle aspect ratio for better flowability. And also it can tolerate higher moisture level. However, the cutting tools in this machine experience significant wear when processing dirty feedstocks. Even the target tool life is 12,000 hours. In reality sometimes as few as 200 hours. So therefore, in this study, we want to use the material technology to understand the problem and propose new design and a new surface treatment and new materials to mitigate the wear issue.
The first thing we need to understand is the wear mode experienced by the Crumbler rotary shear is very different from the conventional milling machine. For example, with a hammer mill, it rotates at very high speed, so therefore the dominant wear mode was erosion. And for the knife mill is a combination of cutting and crushing, so therefore both abrasion and erosion wear were observed. For the Crumbler rotor shear, since the cutting tool rotates at a much lower speed, and therefore the majority of wear is caused by abrasion. The erosion is negligible.
We recently acquired a state-of-the-art high-speed camera to monitor the interface between the cutters and the wood chips to understand the interfacial phenomenon. And you can see from short video clips to the right and the high-speed camera was able to review the moments how wood chips got crushed and sheared between the two sets of cutters. And that allows us to guide more fundamental understanding of the wood chip cutting process and also how the cutting edge of the cutters got worn out.
In the field operation of the rotary sheer, it was found the component wear accelerates in processing dirty feedstock. Therefore, it's important to understand what the species of those dirt are trapped on the feedstock. So we tracked them and analyzed them, so we realized they are primarily minerals. For example quartz and albite. And also we used nano-indentation to confirm their hardness are slightly higher compared to the critical components including spacer, coating plate, and the cutters. Then the next thing we started was the particle distribution of those minerals. So the first thing we learned was the size primarily went from a few microns to a few hundred microns. And the majority is in the tens and hundreds of microns.
And secondly, we noticed there is quite a bite of change for the particle size distribution after rotary shear processing. For example, for the size between 70 to 200 micrometers, those mineral particles, the number of those mineral particles was reduced quite a bit from 21% to 6% after the processing. In contrast, the smaller-size particles, say less than 30 microns, the cost actually increased from 22% to 40%. So what happened? So then after analysis of the cutter assembly, we attribute that to the gap between the two addition cutters. The gap between the cutters is about 100 microns new and up to 200 microns used. So that means during the processing the particle, the group particles in the size of 70 to 200 microns got trapped in between the two cutters and they got crushed so to become smaller particles. But during the crushing process, they create a lot of wear on the side surface of the cutters as well. So that has been confirmed by the picture in the middle here showing that the new cutter versus old cutter. So clearly there is a close correlation between the extrinsic mineral particles trapped in the biomass feedstock and the tool wear we need to study for the rotary shear.
With understanding of the dominant wear mode of the cutters being abrasive wear, so we slightly coated materials for higher resistance to replace the current alloys for the thin and thick cutters. And both baseline candidate materials or testing using the ASTM standard two-body abrasion test, and results are summarized in the chart in the middle of the slide. And clearly, using the candidate D2 or M2 steel, we could potentially improve resistance. And combining the cost factor listed in the tables to the right, we can see the best candidate so far for the thin cutter would be the M2 steel, which can increase the resistance by seven times with the cost of about three and a half times. Now for the thick cutter, the M2 steel would be the better replacement, which doubles the tool life with the cost of 25% increase.
This slide compares the resistance of those candidate coatings testing using the ASTM standard operating test in comparison with the baseline alloys used by the current cutters. And evidently, the iron boriding could reduce the wear rate by a few times up to 20 times with a reasonable cost. Then the thin DLC coating performed even better, with the potential extension of tool life by three orders of magnitude. Then the cost was even lower. The only concern would be the current boriding coating is only a few microns thick. That may or may not last long enough in the actual operation. So that’s why at this moment we are looking into a thicker version of the coating, and hopefully that can surpass the long-term operation need.
Finite element analysis was used to calculate the kind of pressure between the wood chips against the cutters based on the cutter assembly operation and the anisotropic mechanical properties of the wood species. Finite element analysis was used to compare the kind of pressure generated by two cutter designs, as shown to the top red color. The red cutter is the original design, a sharp color of the teeth. Then the green cutter compares the DZ teeth, which is squared collar new design. And these two cutters were simulated to process four different type of wood chips containing different levels of moisture and in different particle sizes. The table shows down there you can see the new cutter evidently experienced lower contact pressure against the wood chips in comparison with the original cutter design. That would expect a lower wear as well. So the reduced wear rate of the new cuter actually has been confirmed in the field experiments.
Based on the bench-scale wear testing and finite element analysis, we are currently working with a cutter manufacturer and coating vendors to fabricate prototype cutters using the new DZ tooth design and also the candidate wear-resistant coatings. And on the bottom of this slide you can see we already produced a D2 tool steel prototype cutters and also applied two coatings of the iron boriding and thin DLC coating onto the both thin and thick cutter surfaces. So we are now—we already shipped those prototype cutters for concept and FCIC is planning to test those prototypes side by side with the baseline cutters, actual rotary shear, to compare the performance and the tool life.
George Fenske, Argonne National Laboratory: This next slide summarizes the approach that we have developed to calculate the recession of the leading edge of the hammer mills and knife mills and so forth. In the case for the hammer mill, the recession is denoted by delta perp as units of meter, and it’s the product of a term that relates the material properties and ash properties to calculate the volume of material eroded per unit mass of ash impinging on the surface times the ash density in terms of kilograms per cubic meter. Similar approach was obtained for knife mills, where we’re using delta Q abrasion where in this case delta Q abrasion is the volume of material, cubic meters per particle that is removed. And in this case chi is the particle density.
And the rotary sheer is a similar approach because we’re using abrasion as delta Q abrasion and the particle density. The difference between the knife mill and the rotary shear are these functions f, which represents the length of the scratches, as well as L, the length of the scratches for a rotary shear system. All of these determinations require information on feedstock attributes, material attributes, as well as process design parameters. All of these parameters enter into these calculations, and you can then come back and determine what the recession is on the leading edges.
This next slide shows the forms taken for delta Q, the volume of material removed either due to erosion or due to abrasion. For the erosion case, delta Q erosion is the sum of two individual units. One is one due to deformation of the surfaces and represents the material removed into the normal component of the velocity coming in while delta Q is sometimes called the cutting portion of it and represents the horizontal energy associated with that. This work adopts the approach that Ben-Ami published around 5 years back and this is the form they’re taking for delta QC and delta QD. You’ll see that in order to calculate the volumetric erosion information on the particle—of the density of the individual particles, the size of the particles, the shapes of the particles, as well as the hardness of the substrate, the toughness of the substrate, the incoming velocity, angle of incidence. And then for the deformation you need the plastic ductility, as well as the shape and hardness of it. So all of these enter into calculating the volumetric loss per kilogram of ash hitting the surface.
In terms of three-body abrasion we went back to the basic derivation of the abrasive wear, in which case we were modeling a particular ash particle. Assuming it’s spherical the diameter D with a load W applied to it, so many [inaudible] applied to it. As it indents into the surface you’ll have an indentation depth called H. And the idea of calculating volume is the volume of this darkened area down here. It’s a cross-section times the length of the scratch. So with that we can have the following relationship—or this information right here is the cross-sectional area of the indentation and then the length of the scratch. The material properties come in when you determine the indentation depth H as well as the particle diameter. Now the indentation depth can either be elastic or plastic in nature. And here that’s where you start to use either the elastic modules or the hardness of the material to determine what the indentation depth is.
In order to validate the erosive wear model, we decided to perform sample calculation using the Ben-Ami wear model and compare those to experimental data obtained using the accelerated wear test rig at Idaho National Laboratory, which is shown in the upper left-hand corner. So we performed a number of calculations, one at 100, 150, 200, and 250 meters per second, compared the results against the experimental data that was adjusted to reflect the kilogram of ash instead of kilogram of biomass—essentially knew what the ash content was. So what you can see in this case is that the experimental data shown in the red dots also agreed quite well with the predicted value either at—and particularly between the 200 and 250 meters per second. So we’re quite happy with the results since we did not change any of the leading coefficients and we used material properties of the actual coupon that were used in these tests.
The next stage in the validation was to compare the predicted results against the wear observed in an actual hammer obtained from the stage one hammer in the Idaho National Lab PDU unit out there. These images on the right show the drum in the hammer mill. It’s about a meter-and-a-half long by about a meter in diameter, and it consists of a number of hammers that are used. Image in the lower, in the middle, shows the hammer. These are typically about 10 inches long, thickness is about 3/8 of an inch. On the right we’re showing the images of a new hammer, which has a sharp edge, as well as the side profile of a worn hammer, which has a rounded edge that was rounded due to erosion on that.
So we apply the QBD process that incorporates the end process primers for the hammer mill, the feedstock properties, as well as the materials of construction, those material properties to try and predict the absolute volumetric wear. But unfortunately the data that provided was lacking. There was no tracing as to how much ash each hammer corner was exposed to, so we went back to a different approach, which is illustrated on the next slide.
So the next stage in validation after validating the results against the experimental accelerated wear test data, we wanted to validate it against a prototypic hammer that was extruded from the Vermeer hammer mill. And the model that we developed is shown here, where we have a bale of corn stover that’s being fed into the Vermeer hammer mill, which consists of a drum about a meter-and-a-half long by a meter in diameter that contains roughly 96 hammers with 6 rows each containing 16 hammers along each row. And so what we want to calculate is this differential volume of material that's worn off, which is the product shown here of delta R times the width of the hammer H times the recession, the amount that's worn off.
And we calculate that volume compared to the actual predicted volume in terms of cubic meters per revolution of a hammer, which is equal to the product of this delta Q for erosion times the amount of material, ash that is exposed to the hammer during one revolution. So if you go through these equations, which I don't have time, you can end up that you can actually calculate what the recessions is as a function of this delta Q erosion times density times radial position R. So that's the summary as to what we use.
Ok. So the previous slide showed that the actual linear recession is dependent upon two parameters, delta Q erosion, as well as the ash density as a function of position. We use several different assumptions for the ash density. This is listed in the table. I don't want to go into detail on that. But we did these assumptions, and the assumptions are plotted over here where we have various uniform distributions, linear distributions, cubic distributions, and finally an exponential distribution where the ash increases exponentially as a function of position. The item that brings all of these together is a boundary condition that the total amount of mass within that drum is equal to the feed rate times the ash content times some ash residence time, in this case 10 seconds.
If you now take those particle density profiles and put it in this particular equation for the recession, we'll see that we get results shown here, which essentially mimic what the ash distribution is. So that's important thing to note on that. In order to compare it with the shape of the hammer that was actually produced, we take that former curve and we kind of twist it and rotate is so that we're plotting the loss along the horizontal axis and the radial position along here. This allows us to compare this position, the radial positions, pictorially with the radial position along a hammer over here. And if you go through and plot these values either the red or the blue ones, you'll see these, particular measured off of the original surface of the hammer, you'll see that you'll get a profile similar to that that's shown in yellow.
So we feel that that's a fairly good representation. You may be able to adjust the shape of that yellow curve by adjusting a few other parameters. But overall we feel that the model is fairly successful at being able to predict the wear. So several potential applications that this model can now be applied to is to look at the impact of the hammer design such as this face angle, the impact on the rotational speed, the impact on material properties, which is what materials you use to fabricate these hammers out of such that you can extend the lifetime, as well as the impact of feedstock variability. If you change the ash content, the ash size and shape, so what impact will that have on the behavior.
So this next slide here shows a few examples of the application of the wear model. First thing shows the impact of material properties if we're going to go to a different alloy or a different material for the hammers, which is improved hardness, toughness, and fatigue properties. What would be the expected change in the wear life of the material, or the amount of material worn off. So we're plotting fracture toughness, anywhere from 5 kilojoule per meters squared up to 50 kilojoules per meter squared, a few different fatigue ductility parameters, as well as different hardness for 8 and 12. The particular alloy used in the hammers that we've been examining is listed in blue right here with the combination of about 4 GPa, 400 [inaudible] hardness or yeah, [inaudible] hardness. Has a plastic strain ductility that's quite high, around 1.0. And from the databases we observed that this material had a toughness around 20 kilojoules per meter squared. So if you want to reduce the wear, you're looking in this region around here.
So you go to a harder material, perhaps a tougher material will reduce the wear. But the question is can you combine all three different combinations in terms of hardness, ductility, and toughness to obtain a material down here. In many cases that may be difficult to achieve. That may be one of the challenges. It also points out that the existing hammer mill alloy that they're using is quite well optimized. And so you really have to get some extreme changes to go better than what they have. Another example is the impact of rotational speed on hammer wear because it's proportional to the velocity squared. Going to higher speed will increase the wear rate and degrade the hammer quicker on that. And then another example is what happens if you change the face angle. You can see that you have to get above 45° or so before you'll see any measurable change on that.
The project pivoted from hammer milling to knife milling about halfway through, and this is the model that was adopted to represent the knife milling. It consists of stationary blades and a rotating hub in the middle. There are blades attached to the rotating hub. And when they rotate, they will form pinch points, a narrow gap that will actually put the biomass material and produce a finer material. Closer up, we're seeing these rotating blades moving up against the stationary knife. And when the material is caught in between, if the gap is appropriate, you'll either cut, shear, or tear, as illustrated here. So as the edge of this particular knife is abraded it becomes duller, and it opens up this gap, which will lead to either cutting, shearing, or tearing.
The first step in calculating abrasive wear is to first calculate what the indentation depth is as a function of variables. In this particular case, calculate H as a function of particle diameter, the load that's applied to the particle, the elastic module of the material, the hardness, and so forth. And so this slide summarizes a couple of calculations. One is for a particle being indented into the cutter, where we can see what the indentation depth is versus the assumed load on each particle. You can see at low loads that's really dominated by elastic deformation, whereas at higher loads you start getting into the plastic deformation. The other part to look at is the particle biomass pair. In this case because the biomass is much softer, the elastic module, the elastic wear indentation is only dominant at real low loads. Once you get above approximately 10−3 newtons, millinewtons, plastic deformation dominates. So that will dominate what the indentation is like.
This is a rather busy slide, but it shows the calculations that are performed to model what the volumetric wear rate is as a function of this abrasive wear model for delta Q, as well as determining the number of particles that are—that this given region sees on that. When you go through all the work you can actually calculate what the recession is by tooth or so forth of the material as being proportional to this wear of an individual particle times the density number of particles per square centimeter. And so you can calculate this recession per revolution of the material. And again, the material properties all come into the calculation of the indentation depth.
So this is somewhat of a repetition of an earlier slide, which shows the approach that we've taken to model the recession of the leading face of different mill components, be it a hammer mill, knife mill or rotary sheer. We can calculate that linear recession per revolution of that component. And we can apply these numbers to predict the recession. And there are several different applications that we're looking at. And one in the terms of biomass conversion, the ability to predict wear and sharpness of the components that will have an impact in terms of maintenance and replacement, timing the maintenance and replacement of the components, predicting the performance. When the particular components change shape, they are dulled. It's going to take a longer time to process materials so the efficiency will drop. So we can use these models to actually predict how that efficiency is going to change. The design of components, how the shape changes it, as well as a selection of materials—can you identify materials that will give you improved lifetimes on that. Techno-economic analysis is an obvious one that we're looking into, as well as the impact of feedstock variability on wear and performance. In addition to the biomass conversion processes there are other applications where studies along this line will benefit the treatment of MSW, design of combination systems for that process. There's significant interest in polymer and plastic recycling now that you can shred and recycle those materials. Battery recycling is another thing, as well as ore processing that we are looking at.
Jun: That's a quick summary. We are currently in collaboration with industry partners to work on various biomass feedstock size reduction systems. To fundamentally understand their wear mechanisms, device tool materials, and designs, fabricate and test prototype high-performance tools, and generate predictive wear models. So as output we are delivering fundamental understanding of the various issues and recommend mitigations for improved economics by increasing the throughput, tool life, and reducing the downtime and parts consumption.
Ed: Thank you very much. I think that concludes the formal part of the presentation, and we do have one question in the chat. George, I believe this might be for you. George, can you hear me? Jun can you hear me?
George: I cannot see the question though.
Ed: Ok. I'll read it. Let me. In the hammer mill evaluation were the hammers fixed or allowed to swing from the modeling perspective?
George: The modeling at this point assumes that they do not swing back and forth. If they were to swing back and forth, you'd probably have to introduce some distribution as to the amount that the material hammer swings back and forth. And what that would probably do is later would smooth out the, it would round it off even more. In fact that was one of the things I was contemplating for looking at why the predicted yellow line in that curve did not agree specifically. But a lot of it is that the comparison that we're showing there for the predicted shape, you needed to put in input as the number of revolutions. I just assumed a 24-hour run, which was 1.3 million revolutions. And that's the number that comes up. If you were to change that you would see differences in that.
So there is room to play with that. Having a distribution as a hammer swings back and forth. You could introduce that into it. The other thing that the model does not include is that it assumes that that angle fee stays constant. It's obvious that as it wears the angle of impact towards the outer radius is actually changing, and so you would actually set up a—perhaps a time-dependent number for the angle of incidence on that. But going back, no. It did not. It assumed that it was constant. But there are ways to actually include the impact of the hammer swinging back and forth. That's it.
Ed: Thanks, George. And let's see. The question came in. Research so far has been focused on woody feedstocks, so how would herbaceous feedstocks compare? And would that be from the angle of different material attributes on the weigh-in?
George: The herbaceous—well Jun can take that, or I can take it. The herbaceous material is probably going to have different distributions in terms of the feedstock variability, the size and shape. That's one thing. In terms of how we would model that, I think for the hammer mill I could see some impact coming in the rounding and how that impacts it. How to handle it mathematically I'd have to look into that because we're really not entering in for the wear calculations any fracture strength of the herbaceous material. So that's where that would come in. You could perhaps go back and figure out based upon how much wear the hammer occurs, what the impact forces would be, the impact of the curvature would have on the impact forces. And you could do some comparisons with that downline. But the knife mill would probably have a more important impact on that. Jun do you have any comments on that?
Jun: Yeah, yeah. I mean for the experimental side, yes. I'm fully aware of the question. When we change the feedstock, the wear situation is going to change as well. But we do believe that the wear mechanism will stay the same because the wear rate, how fast they wear out, would be different. Then right now for the bench-scale testing for screening, the material solutions we are using the minerals directly. So that's like 100% minerals for the bench-scale operating test. For the blasting test, they did use the very fine wood chips. So yeah. But the main thing is after the screening test, we identified a candidate of material solutions, coatings that were going to fabricate the prototypes that that would be tested in the real machines like the rotary shear or the knife mill. That's planned in the next quarter, next year.
In the AOP, where I'm working right now for next year, one important thing is to understand the impact of the variability of feedstock on the performance of the candidate tools. So yes. I think that would be nice. We can not only operate with the wood chips, but also with herbaceous materials that we can say not only how well the solutions can mitigate wear from the woody material, but also from other feedstocks. So I think that's one of the questions from the peer-review comments as well. So that's being considered and written into the AOP for next year.
Ed: Thanks Jun. Thanks George. We're coming to the close of our hour. I don't see any more questions in the chat. So if there are other questions, please reach out to us. Again we are the Feedstock-Conversion Interface Consortium, www.energy.gov/fcic. You can see our website, get our contact information there. I'd like to thank our presenters and thank everybody for joining us. And please have a safe and healthy rest of the day and a peaceful weekend. Thanks everybody.
Jun: Thank you.
George: Thank you.
[End of Audio]