<?xml version="1.0" encoding="utf-8"?><feed xmlns="http://www.w3.org/2005/Atom" ><generator uri="https://jekyllrb.com/" version="3.10.0">Jekyll</generator><link href="https://controlledmold.com/feed.xml" rel="self" type="application/atom+xml" /><link href="https://controlledmold.com/" rel="alternate" type="text/html" /><updated>2025-09-30T14:45:49+00:00</updated><id>https://controlledmold.com/feed.xml</id><title type="html">Controlled Mold</title><subtitle>A blog about fermentation, engineering, and cell-ag</subtitle><entry><title type="html">Industrial computers (PLCs) for koji, natto, etc.</title><link href="https://controlledmold.com/industrial-computers-for-koji/" rel="alternate" type="text/html" title="Industrial computers (PLCs) for koji, natto, etc." /><published>2023-08-30T00:00:00+00:00</published><updated>2023-08-30T00:00:00+00:00</updated><id>https://controlledmold.com/industrial-computers-for-koji</id><content type="html" xml:base="https://controlledmold.com/industrial-computers-for-koji/"><![CDATA[<p>From time to time, I’ll hear about someone putting together an incubation room for growing koji. By now, most folks have realized that importing Japanese equipment is either too costly or too complicated.</p>

<p>And so, the only viable option is to make one yourself. We already covered the basic dimensions and thermal design of the room, so now let’s talk about our favourite part, the controls - using a PLC.</p>

<p>This is part 3 of “How koji is produced on industrial scales”.</p>

<p>(Click here for <a href="https://controlledmold.com/design-of-the-koji-room-koji-muro/">Part 2 - Design of the koji room</a>)</p>

<p>(Click here for <a href="https://controlledmold.com/industrial-koji-fermentation-equipment/">Part 1 - How koji is produced on industrial scales</a>).</p>

<p>In short, I don’t have the space to build a full size koji room, so we’ll do the next best thing, and build the control panel. It doesn’t really matter what size the actual koji room is - the control system will be the same.</p>

<p><img src="/assets/images/koji_plc1/Controlpanel.png" alt="" width="450" class="center" /></p>

<h2 id="what-is-a-plc-and-why-use-one">What is a PLC, and why use one?</h2>

<p>Industry uses specialized computers called Programmable Logic Controllers, or PLCs, to control and automate all of their manufacturing and processes. These PLCs read and write signals to equipment and other pieces of instrumentation in the field (I/O) and can communicate to other PLCs and Supervisory Control and Data Acquisition (SCADA) systems.</p>

<p>On the surface, PLCs seem completely overpriced for their specs.</p>

<p>Consider this - recently I helped build a PC for a buddy of mine. The main components were:</p>

<ol>
  <li>CPU: AMD Ryzen 5600X with 6 cores, 3.7GHz base clock boosting to 4.6GHz - around <strong>$200 CAD</strong></li>
  <li>GPU: AMD RX 5600XT - around <strong>$400 CAD</strong></li>
  <li>Windows 11 - around <strong>$100 CAD</strong></li>
</ol>

<p><img src="/assets/images/koji_plc1/Gamingpc.JPG" alt="" width="450" class="center" /></p>

<p>This thing can run engineering simulations, design software, play many games at high settings 1440p/144FPS, and more. All in, with the motherboard, chassis, RAM, power supply, etc… this system cost around <strong><u>$1200 CAD</u></strong>.</p>

<p>Compare this with your typical small PLC. We’ll use the North America’s Allen Bradley as an example:</p>

<ol>
  <li>CPU: Allen Bradley CompactLogix 1769-L33ER - around <strong>$4000 CAD</strong></li>
  <li>I/O Modules: 16x Digital Inputs, 16x Digital Outputs, 8x Analog Inputs, 4x Analog Ouputs - around <strong>$3000 CAD</strong></li>
  <li>Allen Bradley Programming Software - around <strong>$1200 CAD</strong></li>
</ol>

<p><img src="/assets/images/koji_plc1/Compactlogix.png" alt="" width="450" class="center" /></p>

<p>This thing can… switch stuff on and off? Take some measurements from sensors? All in, with power supplies, an electrical enclosure, terminal blocks, relays, pushbuttons, and communications, our example industrial panel will easily cost over <strong><u>$15000 CAD</u></strong>.</p>

<p>Why not just use a normal PC? Well, when you’re running a facility where downtime costs $1M per day, the last thing you want is for the production line to be brought down by a 100 dollar component. PLCs need to be rugged, dust, shock, and vibration-proof, often running non-stop for decades on end.</p>

<p>It’s not uncommon to show up to a wastewater treatment plant, open a control panel and find an old Allen Bradley SLC-500 installed in the early 90s still chugging along. For this kind of application (municipal infrastructure), you need something with that kind of reliability. The same goes for refineries, manufacturing plants, food processing plants, etc.</p>

<p>PLCs are also designed based on well-known industrial standards, making them easily compatible with industrial instrumentation and programming requirements. Examples:</p>

<ol>
  <li>Want to measure <strong>temperature</strong>? Get a thermocouple (TC) or resistance temperature detector (RTD) input card, and just select the correct TC / RTD type in your PLC program. Or, use a TC/RTD transmitter for a 4-20mA analog input.</li>
  <li>Want to measure <strong>flow</strong>? Communicate from the flow meter via 4-20mA to your analog input. The flow meter will also come with outputs for fault status which you can wire to your digital inputs. You can even get additional diagnostics via the HART protocol.</li>
  <li>Want to implement a Proportional-Integral-Derivative (<strong>PID</strong>) controller? All PLCs come with this function block in software - just drag-and-drop it in, set your tuning parameters (K_p, K_i, and K_d) and you’re done.</li>
</ol>

<p>Of course, you can still do these things with something like an Arduino or Raspberry Pi, but you would need some custom circuitry (like a HAT - “Hardware Attached on Top”) and programming to do so. PLCs, on the other hand, can be serviced quickly by standardized, off-the-shelf components and  programmed by any instrumentation / PLC technician.</p>

<p>In summary, we use PLCs because they are:</p>

<ol>
  <li>Extremely rugged</li>
  <li>Customizable, modular and easily repaired or replaced</li>
  <li>Serviceable and programmable by any competent technician</li>
</ol>

<p>However, PLCs are no more than simple, real-time computing systems. Your average PLC is probably weaker in raw compute power than a Raspberry Pi, and multiple times the cost.</p>

<div class="callout">
💡 Note: We'll still be using a Raspberry Pi as the computer for SCADA. More on that in a later post. Eventually, most PLCs need to communicate back to a "normal" computer via Ethernet/IP. In this case, the Rasp Pi can fail and the PLC will keep running, so reliability isn't an issue.
</div>

<h2 id="why-not-just-use-a-temperature-and-humidity-controller">Why not just use a temperature and humidity controller?</h2>

<p>If you’re just designing an incubator that just needs to be kept at a steady temperature and humidity, this is fine: save some money and use a temperature controller and humidity controller.</p>

<p>Most home koji-makers are already familiar with Inkbird’s temperature and humidity controllers.</p>

<p>These devices are nice and simple. For a temperature controller, you just set a temperature setpoint (SP) and a deadband. If the measured temperature (PV) is below the setpoint minus deadband, you turn a heating element ON via a relay. Once temperature reaches the setpoint, the heating element turns off.</p>

<p>This works really well for home fermentation, and I highly recommend Inkbird’s products, or an equivalent panel-mounted temperature controller.</p>

<p>Thank you, Inkbird (they are not a sponsor of this post).</p>

<p>But what if things get more complicated? What if you need to measure the product (e.g. koji) temperature, the room temperature, AND the relative humidity, and you have some complicated time-dependent logic for all three? What if you need to use a heating element with a variable output on a PID loop? And then add a CO2 sensor with ventilation controls? And then add a load cell to measure live product weight?</p>

<p>Take the standard natto production graph, for example. The room temperature, humidity, and product temperature setpoints are constantly changing as a function of time.</p>

<p><img src="/assets/images/koji_plc1/Nattograph.jpg" alt="" width="450" class="center" /></p>

<p>Not something you can pull of with a simple controller without sitting in front of it and changing the setpoint constantly.</p>

<h2 id="building-a-plc-system">Building a PLC system</h2>

<p>Hopefully you’re convinced by now, so let’s put together a PLC system.</p>

<h3 id="inputs-and-outputs">Inputs and Outputs</h3>

<p>Our first system will be simple. The PLC will have the following I/O:</p>

<ol>
  <li><strong>Analog Inputs (AI):</strong> Room RH (4-20mA), Room CO2 (4-20mA), Room temperature (Type K Thermocouple), and Product Temperature (Type K Thermocouple).</li>
  <li><strong>Analog Outputs (AO):</strong> For now, none. In the future, we’ll add one AO from a PID loop to control a variable heating element, and possibly variable-rate ventilation.</li>
  <li><strong>Digital Inputs (DI):</strong> For now, none. In the future, we can add some basic pushbuttons and selector switches that let the user reset the system, stop everything, and select between LOCAL (panel control) and REMOTE (SCADA control).</li>
  <li><strong>Digital Outputs</strong> <strong>(DO)</strong>: We’ll wire digital outputs to relays to control: Heating, Humidification, Cooling, and Ventilation.</li>
</ol>

<h3 id="hardware-selection">Hardware Selection</h3>

<p>I won’t be forking over $15000 to put together an Allen Bradley system. Thankfully, our koji room is “light industrial” at best, so we can get away with a MUCH cheaper and simpler PLC. In the spirit of making Japanese food, we’ll be using a Japanese PLC line called Koyo Click. Here’s the main components of our build to accommodate our I/O:</p>

<ol>
  <li>Koyo Click C2-01CPU: Click Plus Ethernet - <strong>$97 USD</strong></li>
  <li>Koyo Click C2-08D2-6C: I/O module with 4x DIs, 4x DOs, 4x AIs (4-20mA), and 2x AOs (4-20mA) - <strong>$90 USD</strong></li>
  <li>Koyo Click C0-04THM: Input module with 4x TCs - <strong>$217 USD</strong> (TC/RTD modules are so expensive…)</li>
</ol>

<p><img src="/assets/images/koji_plc1/plc1.JPG" alt="" width="450" class="center" /></p>

<p>Bonus: programming software for Koyo Click is free!</p>

<p>Thank you, Automation Direct (they are not a sponsor of this post).</p>

<p>Since this project is mostly ‘for fun’, we can also also get away with refurbished or used industrial components. Our target budget for this project will be around $1200 CAD - like your average home PC build, and not out of reach for small scale koji/natto/mushroom makers.</p>

<p>After purchasing the necessary DIN rails, breakers, fuses, power supplies, terminal blocks, and relays, we can set up our system on a mock “test bench”:</p>

<p><img src="/assets/images/koji_plc1/testbench1.JPG" alt="" width="450" class="center" /></p>

<p>Here’s where you can really appreciate the convenience of industrial control panels. Everything here is mounted on a DIN rail, a piece of steel channel 35mm wide that allows users to switch out any component as needed. All wires are terminated at screw-in terminal blocks, meaning things can be re-wired quickly without soldering, easily diagnosed with a multimeter, but still be secure enough not to come loose from vibrations.</p>

<p>On the back of the test bench, we’ll mount our sensors. These would normally go into the ‘field’, in our case, the koji room.</p>

<p><img src="/assets/images/koji_plc1/testbench2.JPG" alt="" width="450" class="center" /></p>

<div class="callout">
⚠ We are dealing with 120VAC line voltage, which can be dangerous. If you plan on DIY-ing something based on this post, please make sure you know what you're doing, or get a competent electrician. Also, there are codes that you must follow if you're actually putting together an industrial panel. This post is mostly for 'educational' purposes - my intent isn't to tell you how to put together a panel, but how PLCs work at a basic level. Stay safe!
</div>

<h2 id="next-steps-and-notes">Next Steps and Notes</h2>

<p>That’s it for now! I’ll do some basic testing on this test bench before transferring it to a proper electrical enclosure, with some additional functionality.</p>

<p>I’ll be using this PLC to control a “general purpose” solid-state fermentation room, in order to produce time-dependent temperature/RH/CO2 profiles for koji, natto, and mushroom growing.</p>

<p>This post barely scratches the surface of the capabilities of PLCs. I won’t be going into any wiring details, electrical connections, how relays work, etc. If you want to learn more, see the ‘References’ section below.</p>

<p>A few final notes:</p>

<ul>
  <li>There are plenty of “budget” PLCs out there. I haven’t worked with all of them, so I can’t really offer my opinion on them. Certainly Koyo Click and Productivity Open, both offered by AutomationDirect, are the cheapest.</li>
  <li>Having gone through this exercise, I’m not really sure what the best way of measuring temperature is for koji applications (TCs vs RTDs vs others). I might try multiple methods, just for fun.</li>
  <li>If you’re looking to buy PLC, automation, and control panel components, I found AutomationDirect, RS Components International, and DigiKey to be the best. Looks like McMaster-Carr no longer sells to individuals, and Amazon products are shoddy at best (usually missing CSA/UL labels, certifications, etc.) Used stuff is also a great option!</li>
</ul>

<h2 id="references">References</h2>

<p>In the world of industrial automation, there is a famous, free, 3000+ page PDF by Tony R. Kuphaldt called ‘Lessons In Industrial Instrumentation’ which covers pretty much everything about PLCs and industrial controls.</p>]]></content><author><name>keith</name></author><category term="industry" /><category term="koji" /><category term="solid-state-fermentation" /><category term="scale-up" /><summary type="html"><![CDATA[From time to time, I’ll hear about someone putting together an incubation room for growing koji. By now, most folks have realized that importing Japanese equipment is either too costly or too complicated. And so, the only viable option is to make one yourself. We already covered the basic dimensions and thermal design of the room, so now let’s talk about our favourite part, the controls - using a PLC. This is part 3 of “How koji is produced on industrial scales”. (Click here for Part 2 - Design of the koji room) (Click here for Part 1 - How koji is produced on industrial scales). In short, I don’t have the space to build a full size koji room, so we’ll do the next best thing, and build the control panel. It doesn’t really matter what size the actual koji room is - the control system will be the same. What is a PLC, and why use one? Industry uses specialized computers called Programmable Logic Controllers, or PLCs, to control and automate all of their manufacturing and processes. These PLCs read and write signals to equipment and other pieces of instrumentation in the field (I/O) and can communicate to other PLCs and Supervisory Control and Data Acquisition (SCADA) systems. On the surface, PLCs seem completely overpriced for their specs. Consider this - recently I helped build a PC for a buddy of mine. The main components were: CPU: AMD Ryzen 5600X with 6 cores, 3.7GHz base clock boosting to 4.6GHz - around $200 CAD GPU: AMD RX 5600XT - around $400 CAD Windows 11 - around $100 CAD This thing can run engineering simulations, design software, play many games at high settings 1440p/144FPS, and more. All in, with the motherboard, chassis, RAM, power supply, etc… this system cost around $1200 CAD. Compare this with your typical small PLC. We’ll use the North America’s Allen Bradley as an example: CPU: Allen Bradley CompactLogix 1769-L33ER - around $4000 CAD I/O Modules: 16x Digital Inputs, 16x Digital Outputs, 8x Analog Inputs, 4x Analog Ouputs - around $3000 CAD Allen Bradley Programming Software - around $1200 CAD This thing can… switch stuff on and off? Take some measurements from sensors? All in, with power supplies, an electrical enclosure, terminal blocks, relays, pushbuttons, and communications, our example industrial panel will easily cost over $15000 CAD. Why not just use a normal PC? Well, when you’re running a facility where downtime costs $1M per day, the last thing you want is for the production line to be brought down by a 100 dollar component. PLCs need to be rugged, dust, shock, and vibration-proof, often running non-stop for decades on end. It’s not uncommon to show up to a wastewater treatment plant, open a control panel and find an old Allen Bradley SLC-500 installed in the early 90s still chugging along. For this kind of application (municipal infrastructure), you need something with that kind of reliability. The same goes for refineries, manufacturing plants, food processing plants, etc. PLCs are also designed based on well-known industrial standards, making them easily compatible with industrial instrumentation and programming requirements. Examples: Want to measure temperature? Get a thermocouple (TC) or resistance temperature detector (RTD) input card, and just select the correct TC / RTD type in your PLC program. Or, use a TC/RTD transmitter for a 4-20mA analog input. Want to measure flow? Communicate from the flow meter via 4-20mA to your analog input. The flow meter will also come with outputs for fault status which you can wire to your digital inputs. You can even get additional diagnostics via the HART protocol. Want to implement a Proportional-Integral-Derivative (PID) controller? All PLCs come with this function block in software - just drag-and-drop it in, set your tuning parameters (K_p, K_i, and K_d) and you’re done. Of course, you can still do these things with something like an Arduino or Raspberry Pi, but you would need some custom circuitry (like a HAT - “Hardware Attached on Top”) and programming to do so. PLCs, on the other hand, can be serviced quickly by standardized, off-the-shelf components and programmed by any instrumentation / PLC technician. In summary, we use PLCs because they are: Extremely rugged Customizable, modular and easily repaired or replaced Serviceable and programmable by any competent technician However, PLCs are no more than simple, real-time computing systems. Your average PLC is probably weaker in raw compute power than a Raspberry Pi, and multiple times the cost. 💡 Note: We'll still be using a Raspberry Pi as the computer for SCADA. More on that in a later post. Eventually, most PLCs need to communicate back to a "normal" computer via Ethernet/IP. In this case, the Rasp Pi can fail and the PLC will keep running, so reliability isn't an issue. Why not just use a temperature and humidity controller? If you’re just designing an incubator that just needs to be kept at a steady temperature and humidity, this is fine: save some money and use a temperature controller and humidity controller. Most home koji-makers are already familiar with Inkbird’s temperature and humidity controllers. These devices are nice and simple. For a temperature controller, you just set a temperature setpoint (SP) and a deadband. If the measured temperature (PV) is below the setpoint minus deadband, you turn a heating element ON via a relay. Once temperature reaches the setpoint, the heating element turns off. This works really well for home fermentation, and I highly recommend Inkbird’s products, or an equivalent panel-mounted temperature controller. Thank you, Inkbird (they are not a sponsor of this post). But what if things get more complicated? What if you need to measure the product (e.g. koji) temperature, the room temperature, AND the relative humidity, and you have some complicated time-dependent logic for all three? What if you need to use a heating element with a variable output on a PID loop? And then add a CO2 sensor with ventilation controls? And then add a load cell to measure live product weight? Take the standard natto production graph, for example. The room temperature, humidity, and product temperature setpoints are constantly changing as a function of time. Not something you can pull of with a simple controller without sitting in front of it and changing the setpoint constantly. Building a PLC system Hopefully you’re convinced by now, so let’s put together a PLC system. Inputs and Outputs Our first system will be simple. The PLC will have the following I/O: Analog Inputs (AI): Room RH (4-20mA), Room CO2 (4-20mA), Room temperature (Type K Thermocouple), and Product Temperature (Type K Thermocouple). Analog Outputs (AO): For now, none. In the future, we’ll add one AO from a PID loop to control a variable heating element, and possibly variable-rate ventilation. Digital Inputs (DI): For now, none. In the future, we can add some basic pushbuttons and selector switches that let the user reset the system, stop everything, and select between LOCAL (panel control) and REMOTE (SCADA control). Digital Outputs (DO): We’ll wire digital outputs to relays to control: Heating, Humidification, Cooling, and Ventilation. Hardware Selection I won’t be forking over $15000 to put together an Allen Bradley system. Thankfully, our koji room is “light industrial” at best, so we can get away with a MUCH cheaper and simpler PLC. In the spirit of making Japanese food, we’ll be using a Japanese PLC line called Koyo Click. Here’s the main components of our build to accommodate our I/O: Koyo Click C2-01CPU: Click Plus Ethernet - $97 USD Koyo Click C2-08D2-6C: I/O module with 4x DIs, 4x DOs, 4x AIs (4-20mA), and 2x AOs (4-20mA) - $90 USD Koyo Click C0-04THM: Input module with 4x TCs - $217 USD (TC/RTD modules are so expensive…) Bonus: programming software for Koyo Click is free! Thank you, Automation Direct (they are not a sponsor of this post). Since this project is mostly ‘for fun’, we can also also get away with refurbished or used industrial components. Our target budget for this project will be around $1200 CAD - like your average home PC build, and not out of reach for small scale koji/natto/mushroom makers. After purchasing the necessary DIN rails, breakers, fuses, power supplies, terminal blocks, and relays, we can set up our system on a mock “test bench”: Here’s where you can really appreciate the convenience of industrial control panels. Everything here is mounted on a DIN rail, a piece of steel channel 35mm wide that allows users to switch out any component as needed. All wires are terminated at screw-in terminal blocks, meaning things can be re-wired quickly without soldering, easily diagnosed with a multimeter, but still be secure enough not to come loose from vibrations. On the back of the test bench, we’ll mount our sensors. These would normally go into the ‘field’, in our case, the koji room. ⚠ We are dealing with 120VAC line voltage, which can be dangerous. If you plan on DIY-ing something based on this post, please make sure you know what you're doing, or get a competent electrician. Also, there are codes that you must follow if you're actually putting together an industrial panel. This post is mostly for 'educational' purposes - my intent isn't to tell you how to put together a panel, but how PLCs work at a basic level. Stay safe! Next Steps and Notes That’s it for now! I’ll do some basic testing on this test bench before transferring it to a proper electrical enclosure, with some additional functionality. I’ll be using this PLC to control a “general purpose” solid-state fermentation room, in order to produce time-dependent temperature/RH/CO2 profiles for koji, natto, and mushroom growing. This post barely scratches the surface of the capabilities of PLCs. I won’t be going into any wiring details, electrical connections, how relays work, etc. If you want to learn more, see the ‘References’ section below. A few final notes: There are plenty of “budget” PLCs out there. I haven’t worked with all of them, so I can’t really offer my opinion on them. Certainly Koyo Click and Productivity Open, both offered by AutomationDirect, are the cheapest. Having gone through this exercise, I’m not really sure what the best way of measuring temperature is for koji applications (TCs vs RTDs vs others). I might try multiple methods, just for fun. If you’re looking to buy PLC, automation, and control panel components, I found AutomationDirect, RS Components International, and DigiKey to be the best. Looks like McMaster-Carr no longer sells to individuals, and Amazon products are shoddy at best (usually missing CSA/UL labels, certifications, etc.) Used stuff is also a great option! References In the world of industrial automation, there is a famous, free, 3000+ page PDF by Tony R. Kuphaldt called ‘Lessons In Industrial Instrumentation’ which covers pretty much everything about PLCs and industrial controls.]]></summary><media:thumbnail xmlns:media="http://search.yahoo.com/mrss/" url="https://controlledmold.com/assets/images/koji_plc1/cover.JPG" /><media:content medium="image" url="https://controlledmold.com/assets/images/koji_plc1/cover.JPG" xmlns:media="http://search.yahoo.com/mrss/" /></entry><entry><title type="html">I wrote a book on Natto</title><link href="https://controlledmold.com/natto/" rel="alternate" type="text/html" title="I wrote a book on Natto" /><published>2023-04-11T00:00:00+00:00</published><updated>2023-04-11T00:00:00+00:00</updated><id>https://controlledmold.com/natto</id><content type="html" xml:base="https://controlledmold.com/natto/"><![CDATA[<p>It was fun to write. Here are a few thoughts on natto, writing, and academia…</p>

<h3 id="there-were-no-detailed-books-on-natto">There were no detailed books on Natto</h3>

<p>In 2022, if you searched ‘natto book’ on Amazon.com, the first result is a self-help / insomia book: something about a problem getting stickier the more you stir it. I’m sure it’s a great book, but… come on.</p>

<p>Aside from that, there are a few decent home fermentation books with passing mentions about natto. Kristen and Christopher Shockey’s book is great, and so is Sandor Katz’s. Natto chapters also show up in a few textbooks on fermented foods, but none of them have the sort of detail I’m looking for.</p>

<p>What kind of book did I want? Something like this:</p>

<p><img src="/assets/images/natto_book/kojistudies_book.JPG" alt="" width="550" class="center" /></p>

<p>I’ve torn apart quite a few of these by now. The one above is Koji Studies, published by the Brewing Society of Japan. Their books are really good: technical, but practical. They have whole chapters on history, industrial processes, biology, and biochemistry. They’re geared toward industry operators and researchers, but I like that a bit better than general audience books.</p>

<h3 id="translating-japanese-textbooks">Translating Japanese Textbooks</h3>

<p>There wasn’t a complete technical natto book in English, so I had to write one. Thankfully, there are some good Japanese natto books, so I bought them all, and transcribed and translated them:</p>

<p><img src="/assets/images/natto_book/nattoref_book.JPG" alt="" width="550" class="center" /></p>

<p>This was easy enough. After scanning all those Brewing Society of Japan books, natto books were a breeze. Combined with a giant literature review on natto and <em>Bacillus subtilis</em>, I had enough to write an English-language book on natto. I tried to write with a decent mix of technical info, without getting impractical. Here’s a preview:</p>

<p><img src="/assets/images/natto_book/preview.JPG" alt="" width="550" class="center" /></p>

<p>Here are a few fascinating things I learned about natto, and a few thoughts.</p>

<h3 id="things-i-learned">Things I learned</h3>

<h4 id="i-thought-natto-and-bacillus-subtilis-would-be-boring">I thought natto and <em>Bacillus subtilis</em> would be boring</h4>

<p>After all, it’s just bacteria on beans, right? Fungi like seem way more interesting, with all the hyphae, fruiting bodies, and specialized methods of reproduction. I was wrong, and I’m now convinced that every single organism on Earth is interesting.</p>

<p><em>B. subtilis</em> can: form hardy endospores (gram-negatives like <em>E. coli</em> can’t do this) that survive boiling water, swim, swarm, form multicellular communities with aerial ‘wrinkles’, exchange DNA, commit fratricide, produce a detergent-like substance that breaks the surface tension of water… and many, many others.</p>

<p>It’s crazy how complicated something as innocuous as bacteria can be.</p>

<h4 id="there-are-a-lot-of-natto-types-out-there">There are a LOT of natto types out there</h4>

<p>Different sized beans, red soybeans, black soybeans, green soybeans, cracked soybeans, different sauces, you name it. There’s a huge diversity in natto products in Japan, kind of like our yogurt and cheese sections in the West.</p>

<p>There are also different strains of <em>Bacillus subtilis</em> used for natto, which give properties like low-odour and low-stickiness natto. These were two of the more interesting sections I wrote about, and they’re the most promising for natto first-timers.</p>

<p>Aside from that, I hope people try making natto with species other than <em>B. subtilis</em>, or mixed cultures of different bacteria. I think there’s a lot of potential for whole soybean foods and ferments in the future.</p>

<h3 id="thoughts-on-academia">Thoughts on Academia</h3>

<h4 id="low-quality-papers">Low Quality Papers</h4>

<p>Without naming specifics, I came across some journals with really questionable content. For some papers, it seemed like the writers weren’t even trying when it came to things like fact-checking or putting together decent diagrams (MS Paint, anyone?).</p>

<p>It does feel like academia in the last decade has partially become a mill to churn out papers. Someone should do a tier list of academic journals. Respect to those watching out for, and speaking up about academic integrity.</p>

<h4 id="medical-studies">Medical Studies</h4>

<p>Reading medical studies is a headache. There are so many issues to watch out for, like conflicts of interest, trials that are too short, photoshopped results (shoutout to Elisabeth Bik), playing with statistics, etc. And then there are the media outlets making bold claims and headlines over inconclusive studies.</p>

<p>As someone not in the medical field, these papers can get really hard to read. I was definitely in over my head when following a decades-long academic rabbit hole on what all the Vitamin K-dependent proteins actually do. Short answer: we’re still not sure.</p>

<h3 id="second-edition-and-other-topics">Second Edition and Other Topics</h3>

<p>Nevertheless, I enjoyed self-publishing: it allowed me to learn about things like cover design, graphic design, formatting, front matter, ISBNs, and publishing in general. It was fun to write, draw diagrams, and do photoshoots.</p>

<p>My only regret is that self-published books in Canada no longer get a Cataloguing in Publication (CIP) data block, something I wanted which would make it easier to get my book into a library. I’ll probably try to write and publish a Second Edition through a traditional publisher, and add more product photos and some microscopy shots.</p>

<p>If I write again, I think my next book would be much faster, now that I (sort of) know what I’m doing. Some other topics I’d like to write about are:</p>

<ol>
  <li><strong>Koji</strong> - but there are a ton of decent koji books now, and good websites (Check out <a href="http://www.kojiwiki.com/index.php/Main_Page">Koji Wiki</a>!)</li>
  <li><strong>Space Food</strong> - this might have to start as a series of blog posts, first. I also like the idea of minimalist food, dystopian food, food ideas for the future, etc.</li>
</ol>

<h3 id="where-to-buy">Where to buy</h3>

<p>You’ll be able to buy my book, <strong>Natto: History, Science and Technology</strong>, in hardcover or paperback on Amazon or Barnes &amp; Noble by searching the title.</p>

<p>E-book will follow later this winter.</p>]]></content><author><name>keith</name></author><category term="industry" /><category term="natto" /><category term="solid-state-fermentation" /><summary type="html"><![CDATA[It was fun to write. Here are a few thoughts on natto, writing, and academia…]]></summary><media:thumbnail xmlns:media="http://search.yahoo.com/mrss/" url="https://controlledmold.com/assets/images/natto_book/cover.JPG" /><media:content medium="image" url="https://controlledmold.com/assets/images/natto_book/cover.JPG" xmlns:media="http://search.yahoo.com/mrss/" /></entry><entry><title type="html">Quorn™: a story about Single Cell Protein</title><link href="https://controlledmold.com/quorn-a-story-about-single-cell-protein/" rel="alternate" type="text/html" title="Quorn™: a story about Single Cell Protein" /><published>2022-01-14T00:00:00+00:00</published><updated>2022-01-14T00:00:00+00:00</updated><id>https://controlledmold.com/quorn-a-story-about-single-cell-protein</id><content type="html" xml:base="https://controlledmold.com/quorn-a-story-about-single-cell-protein/"><![CDATA[<p>What if you could make food out of just basic chemicals and microbes? At a first glance, this seems like a great idea since microrganisms grow quickly, are thermodynamically efficient, and can add value to low-cost chemicals and waste streams.</p>

<p>This concept, which gained a lot of traction in the 1960s and 1970s, is called “Single Cell Protein (SCP)”, in which a microbe (usually fungi, bacteria, or algae) is cultured for direct consumption. Although “microbial biomass” might be a more accurate term, we’ll continue to use the term SCP.</p>

<p>This post is about one SCP product: <em>Quorn</em> - which is the brand name for imitation meats based on cultured <em>Fusarium venenatum</em>, a soil mold that has become one of a few successful SCP products to date. And while far from perfect, it’s still a fascinating story.</p>

<h3 id="the-inception-of-quorn-and-the-protein-crisis">The Inception of Quorn and the ‘Protein Crisis’</h3>

<p>Between the 1950s and 1970s, amidst a massive global population boom, there was a concern that children in developing countries were becoming malnourished due to a lack of protein. This culminated in a 1968 paper from the United Nations with policies to avert this ‘protein crisis’, and one of their recommendations was to promote the development of single-cell protein for direct consumption or animal feeding [1]. This sounds promising at first: microorganisms grow fast, don’t take up a lot of space, and are <strong>30% - 80% protein</strong> by dry mass. [2]</p>

<p>Around 1967, the British company Rank Hovis McDougall (RHM) took on the challenge, eventually selecting a strain of <strong><em>Fusarium venenatum</em></strong> (A 3/5) out of 3000+ candidates. To scale-up, RHM initially worked with DuPont using continuously stirred tank reactors (CSTRs) before collaborating with the Imperial Chemical Industries (ICI) to use their airlift bioreactor - more on that later [3]. This joint venture would be known as Marlow Foods.</p>

<p>There are plenty of articles highlighting the history of Quorn, which I’ve linked in the References section. Although the history and biology of Quorn is interesting, I want to focus on two things in this post:</p>

<ol>
  <li>How did Quorn succeed at scale, from an engineering perspective?</li>
  <li>What did Quorn actually accomplish?</li>
</ol>

<h3 id="bioreactor-design---upstream-processing">Bioreactor Design - Upstream Processing</h3>

<p>Growing <em>F. venenatum</em> in a liquid culture is fairly straightforward. Like other liquid cultures, the fungi can be propogated in a sterile culture of carbohydrates, ammonia, oxygen, and nutrients: yielding biomass and carbon dioxide.</p>

<p>This can be accomplished with any bioreactor design, but after two iterations, the design team eventually opted for an <strong>external-loop airlift bioreactor</strong> [4-7] that looks a bit like a Klein bottle from mathematics:</p>

<p><img src="/assets/images/quorn/airliftbioreactor.png" alt="" width="550" class="center" /></p>

<p>Here’s how it works:</p>

<ol>
  <li>Sterile water, carbohydrates (as syrup or molasses), and nutrients are pumped into the bottom of the bioreactor</li>
  <li>Sterile oxygen (as air) and a nitrogen source (ammonia) are bubbled into the bottom of the bioreactor</li>
  <li>The aerated mixture rises due to its lower average density</li>
  <li>Metabolic heat is removed using cooling water coils</li>
  <li>Waste carbon dioxide and gases are vented at the top of the bioreactor</li>
  <li>The culture broth, now with a compartively higher average density, drops into the downcomer</li>
  <li>A portion of the biomass is harvested and the remainder returns to the bottom of the bioreactor to repeat the cycle</li>
</ol>

<p>As a side-note, the bioreactor design shown above is based off of Quorn-2 and Quorn-3, Marlow Foods’ second and third bioreactor, respectively. The first pilot bioreactor, Quorn-1, was initially used to develop ICI’s Pruteen™ - a bacterial SCP meant for livestock - and looks a bit different. Quorn-2 and Quorn-3, at 150,000L, are the biggest bioreactors of their type.</p>

<p>This is a great design for several reasons:</p>

<ol>
  <li>At the physical scale Quorn is dealing with, the design is more efficient AND gentler than CSTRs, whose rotating blades would damage the growing hyphae. Mixing by aeration only is gentle and suitable for viscous broths with lots of hyphae.</li>
  <li>The loop design avoids <strong>short-circuiting</strong> of fresh carbohydrates and nitrogen from the inlet directly to the outlet, which can occur in conventional CSTRs.</li>
  <li>Adding air to the bottom of a tall bioreactor allows oxygen to dissolve quickly into the media through the immense <strong>hydrostatic pressure</strong></li>
</ol>

<p>The bioreactor is operated <strong>continuously</strong>, typically for several weeks before resetting the bioreactor (for reasons discussed later). Continuously operated bioreactors will always be more productive than batch bioreactors because they maintain a high biomass concentration at all times.</p>

<h3 id="quorn-quirks---downstream-processing">Quorn Quirks - Downstream Processing</h3>

<p>If everything’s gone well, coming out of our bioreactor is a solution of hyphae and spent substrate that’s around 1.5% biomass by weight. A few downstream processing steps are required before the biomass is converted to Quorn mycoprotein.</p>

<h5 id="nucleic-acids">Nucleic Acids</h5>

<p>Things that grow fast and produce proteins quickly contain a lot of RNA for translation. This is a problem because humans metabolize RNA into uric acid and orotic acid, and the overaccumulation these can lead to gout and liver damage, respectively.</p>

<div class="callout">
💡 First note: In humans, purines are metabolized into uric acid, and pyramidines are metabolized into orotic acid. The uric acid problem is primarily related to primates: somewhere on the evolutionary calendar, a mutation in a common ancestor knocked out the gene for uricase in humans and other primates. In non-primates, uric acid is simply metabolized into allantoin, which is soluble and excreted [11]. 
</div>

<p>The <em>F. venenatum</em> biomass has a nucleic acid content of around 8-9% w/w dry basis which is too high for human consumption. To solve this problem, the biomass is heated to 68°C for 15+ minutes, and endogenous RNAases break down cellular RNA into 5’-ribonucleotides, which diffuse out of the cell.  This heating process causes some components of the hyphae to leak out into the surroundings - and up  30% of the biomass is also lost in the process.</p>

<p><img src="/assets/images/quorn/RNAase.png" alt="" width="450" class="center" /></p>

<p>Luckily, some 5’-ribonucleotides are potent umami compounds and synergistic with glutamate. The producers of Quorn concentrate this liquid and sell it as a flavour enhancer.</p>

<h5 id="mutations">Mutations</h5>

<p>Growing a continuous culture over an extended period of time presents an interesting issue: over time, the fungi is susceptible to mutations. Like clockwork, a highly branched mutant of <em>F. venenatum</em> will appear after around 100 generations of growth. The branched mutant grows better than the parent strain in the continuous culture, and will eventually replace it.</p>

<p>This isn’t actually a problem for biomass production, but the branched mutation has an inferior texture to the more linear parent. The goal of Quorn is to simulate meat-like fibres which require linear hyphae.</p>

<p><img src="/assets/images/quorn/Linearvsbranched.png" alt="" width="350" class="center" /></p>

<p>And so, at around 1000h (6 weeks) of operation, the reactor is shut down to be re-seeded with a new culture. For all practical purposes, this is good enough: resetting the reactor at such long intervals is an insignificant loss in productivity and the operators would need a window to clean and perform maintenance anyways.</p>

<div class="callout">
💡 Second note: the branched mutation issue also shows up consistently in liquid cultures of <i>Aspergillus nidulans</i>, <i>A. niger</i> and <i>A. oryzae</i>. It's easy to understand why: filamentous fungi evolved to grow in relatively low-moisture environments, extending their growing hyphae tip deep into a substrate to search for food. If you're in a liquid culture with no competition and glucose everywhere... well, why bother? Just extend hyphae everywhere, as fast as possible, and you'll grow faster than the "normal" parents.
</div>

<h5 id="final-form">Final Form</h5>

<p>The biomass broth, after heating to 68°C to remove nucleic acids (as a bonus: proteases are inactivated), is heated again to 90°C to pasteurize and then centrifuged to yield a 20-30% solids paste. To make Quorn products, the biomass has to be mixed with calcium, egg albumin (vegetarian products) or gluten, carrageenan, and alginates (vegan products). The product is finally frozen, which also improves the texture.</p>

<h3 id="what-did-quorn-achieve">What did Quorn achieve?</h3>

<p>I’m going to offer a few of my own thoughts and speculation on Quorn, and the future of SCP in general.</p>

<h5 id="quorn-as-a-benchmark">Quorn as a Benchmark</h5>

<p>When I started researching Quorn, my primary goal was to develop a mass &amp; energy balance of the overall process to serve as a benchmark. The key driver is that Quorn is a commercially successful product that has been implemented <em>at scale</em> and with a <em>continuously operated bioreactor</em>.</p>

<p>As I write, Quorn is pretty much the only commercially viable SCP product that can be a primary source of protein for humans. Other products like <em>Nutritional Yeast</em>, <em>Yeast Extract</em> and <em>Spirullina</em> don’t really count - they’re more like flavouring agents or supplements.</p>

<p>Any new or competing product better be more efficient, taste better, or be cheaper to produce. The Quorn benchmark becomes useful once we start looking at the economics of newer SCP products and technologies like cultivated meat from animal cells.</p>

<p>Although it was difficult to piece together the mass balance, since process details tend to only be available to plant operators and engineers, I could piece together a decent estimate by combining several sources [4-10] and my own assumptions:</p>

<p><img src="/assets/images/quorn/quornmassbalance.png" alt="" width="550" class="center" /></p>

<h5 id="process-efficiency">Process Efficiency</h5>

<p>So how does Quorn compare to traditional proteins? The “efficiency” of a source of protein for consumption can be measured in two ways:  <strong>caloric</strong> and <strong>protein</strong> conversion efficiency. The caloric conversion efficiency is the ratio of calories output by calorie input (recall: we are adding glucose, a high-calorie feed, to the bioreactor). Similarly, the protein conversion efficiency measures is the ratio of protein output by protein input (animals, like humans, require protein to grow and produce their own protein).</p>

<p>At a high level, we ended up with a bioprocess that is more efficient than animal rearing, cost and nutritionally competitive as a meat substitute, but still more expensive than plant proteins. If we examine the caloric and protein conversion efficiencies of Quorn next to animal proteins [13], we can see that it holds up well enough:</p>

<table class="table-clean">
  <thead>
    <tr>
      <th>Product</th>
      <th>Caloric Conversion Efficiency</th>
      <th>Protein Conversion Efficiency</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>Quorn</td>
      <td>21%</td>
      <td>N/D</td>
    </tr>
    <tr>
      <td>Eggs (Chicken)</td>
      <td>17%</td>
      <td>31%</td>
    </tr>
    <tr>
      <td>Milk (Dairy Cows)</td>
      <td>17%</td>
      <td>14%</td>
    </tr>
    <tr>
      <td>Poultry</td>
      <td>13%</td>
      <td>21%</td>
    </tr>
    <tr>
      <td>Pork</td>
      <td>9%</td>
      <td>9%</td>
    </tr>
    <tr>
      <td>Beef</td>
      <td>3%</td>
      <td>3%</td>
    </tr>
  </tbody>
</table>

<p>The efficiency of Quorn is hampered by the 30% biomass loss at the RNA reduction step - in this table I’ve only considered the centrifuged cake as the consumable end-product. This might be the Achilles’ Heel of all SCP - you can have a microorganism that either <strong>grows quickly</strong>, or has <strong>low nucleic acids</strong> - not both. And removing nucleic acids comes at a hefty cost.</p>

<p>Quorn doesn’t really have a protein conversion efficiency because it directly utilizes ammonia, not protein, as a nitrogen source.</p>

<h5 id="quorn-as-a-pathway">Quorn as a Pathway</h5>

<p>Having a direct pathway from <strong>carbohydrates + ammonia</strong> to <strong>protein</strong> has a lot of neat implications. If plant-based meat substitutes continue to gain traction (and I think they will), the production of plant protein isolates and concentrates is going to yield a lot of waste carbohydrates and starches, which will become dirt cheap. This would make the Quorn process a lucrative pathway to upgrade low-value carbohydrates into protein.</p>

<p>Alternatively, if corn ethanol gets banned due to policy changes, we’re going to have a lot of un-used corn milling and hydrolysing infrastructure. These could easily be adapted into Quorn-type protein production plants.</p>

<h5 id="taste-and-texture-matter">Taste and Texture Matter</h5>

<p>A food products’ <strong>organoleptic properties</strong> matter a lot in the meat substitute space. Quorn’s success comes in part due to the fact the the hyphal product, after heat treatment and freezing, resembles the fibrous texture of meat.</p>

<p><img src="/assets/images/quorn/quorn-meatless-pieces-fried-texture.png" alt="" width="350" class="center" />
<em>Fried Quorn nuggets - Source: veggiepattytastetest.com</em></p>

<h5 id="quorn-as-a-miso">Quorn as a Miso</h5>

<p>I’d love to get my hands on some <em>F. venenatum</em> biomass, centrifuged, <strong>without heat-treating to remove RNA</strong>. I would take this 20-30% solids biomass and heat treat it myself, keeping the 8-9% nucleic acids, and maybe make a miso or garum out of it, since it would contain plenty of natural 5’-ribonucleotides and glutamates. In theory it would make an umami bomb that doesn’t have the flavour of yeast extracts. At the small amounts I would be consuming, uric acid isn’t a problem: so long as daily consumption is under the WHO recommendation of 2g RNA per day (a tablespoon of this stuff should contain around 0.3g).</p>

<p>Maybe I’ll ask Quorn for some, or make my own. The original patents for this bioprocess have already expired [14] - though it would be difficult to replace Marlow Foods’ 35 years of operating experience.</p>

<h5 id="the-protein-crisis">The Protein Crisis</h5>

<p>The original goal at inception, to prevent protein malnutrition, was never really met - because thankfully, and generally speaking - the ‘Protein Crisis’ never happened: agricultural productivity kept up with population growth through the 1960s, 70s, 80s, and beyond. This was mostly due to the <em>Green Revolution</em> consisting of better crop strains, fertilizer use, and the like. SCP could never compete economically with plant proteins.</p>

<h5 id="space-travel-and-the-apocalypse">Space Travel and the Apocalypse</h5>

<p>Despite its limitations, Quorn and SCP in general might be viable technologies in either an apocalyptic scenario - or space travel - as part of a larger system of chemical processes and bioreactors used to sustain human life.</p>

<p>Let’s say an asteroid - on par with the one that wiped out the dinosaurs - impacted earth and a dust cloud shut down photosynthesis for decades, worldwide. That, or we bomb ourselves into a nuclear winter. In either scenario, SCP might be good insurance against a world without photosynthesis. Nuclear, geothermal or wind power would be required to fix carbon and SCP bioreactors might be our only food source once stockpiles run dry.</p>

<p>Or, let’s say we want to colonize space. The same process might be viable as part of a life support system that closes the loop on human sustenance: exhaled CO2, water, and excreted feces and urine are directly recycled into food with energy as an input. This is a huge oversimplification, but at least it’s thermodynamically feasible and may require less space or energy than an equivalent plant-based farm. I’ll explore this in a future post.</p>

<h3 id="conclusion">Conclusion</h3>

<p>Quorn was developed in the 1960s - 1980s in a coordinated academic, institutional and industrial effort, and it survived the test of time even as other SCPs like <em>Toprina,</em> <em>Peliko</em> and <em>Pruteen</em> fell short and were discontinued.</p>

<p>As we enter an era in which meat alternatives are front and centre, Quorn occupies a decent niche - I don’t see it going away any time soon. Recent developments in this space tend to focus on a) animal culture lines or b) food-science based approaches.</p>

<p>That being said, biology and engineering have progressed a lot since the 1960s. We have more species and tools at our disposal, we understand genetics more, and our engineering has improved. If a company were to make the same coordinated academic, institutional, and industrial push for a “Quorn 2.0”, I’m sure it would yield a superior product or process. Here’s hoping.</p>

<p>And as I mentioned earlier, the patent on the original process has already expired [14].</p>

<p>Oh, and I can’t get Quorn in Canada. This might be an issue with the CFIA, or the parent company simply hasn’t made the push into the Canadian market. I’ll pick some up the next time I’m in the USA or UK.</p>

<h3 id="references">References</h3>

<ol>
  <li>Semba, R. D. The Rise and Fall of Protein Malnutrition in Global Health. ANM 2016, 69 (2), 79–88. https://doi.org/10.1159/000449175.</li>
  <li>Ritala, A.; Häkkinen, S. T.; Toivari, M.; Wiebe, M. G. Single Cell Protein—State-of-the-Art, Industrial Landscape and Patents 2001–2016. Frontiers in Microbiology 2017, 8, 2009. https://doi.org/10.3389/fmicb.2017.02009.</li>
  <li>Trinci, A. P. J. Myco-Protein: A Twenty-Year Overnight Success Story. Mycological Research 1992, 96 (1), 1–13. https://doi.org/10.1016/S0953-7562(09)80989-1.</li>
  <li>Finnigan, T. J. A. 13 - Mycoprotein: Origins, Production and Properties. In Handbook of Food Proteins; Phillips, G. O., Williams, P. A., Eds.; Woodhead Publishing Series in Food Science, Technology and Nutrition; Woodhead Publishing, 2011; pp 335–352. https://doi.org/10.1533/9780857093639.335.</li>
  <li>Wiebe, M. G. Quorn Myco-Protein - Overview of a Successful Fungal Product. Mycologist 2004, 18 (1), 17–20. https://doi.org/10.1017/S0269915X04001089.</li>
  <li>Solomons, G. L.; Litchfield, J. H. Single Cell Protein. Critical Reviews in Biotechnology 1983, 1 (1), 21–58. https://doi.org/10.3109/07388558309082578.</li>
  <li>Wiebe, M. Myco-Protein from Fusarium Venenatum: A Well-Established Product for Human Consumption. Appl Microbiol Biotechnol 2002, 58 (4), 421–427. https://doi.org/10.1007/s00253-002-0931-x.</li>
  <li>Ugalde, U. O.; Castrillo, J. I. Single Cell Proteins from Fungi and Yeasts. In Applied Mycology and Biotechnology; Khachatourians, G. G., Arora, D. K., Eds.; Agriculture and Food Production; Elsevier, 2002; Vol. 2, pp 123–149. https://doi.org/10.1016/S1874-5334(02)80008-9.</li>
  <li>Comprehensive Biotechnology. 3: The Practice of Biotechnology: Current Commodity Products, 1st ed.; Blanch, H. W., Moo-Young, M., Eds.; Pergamon Press: Oxford [u.a.] Frankfurt, 1985; Vol. 3.</li>
  <li>Finnigan, T.; Lemon, M.; Allan, B.; Paton, I. Mycoprotein, Life Cycle Analysis and the Food 2030 Challenge. Aspects of Applied Biology 2010, No. No.102, 81–90.</li>
  <li>Johnson, R. J.; Lanaspa, M. A.; Gaucher, E. A. Uric Acid: A Danger Signal from the RNA World That May Have a Role in the Epidemic of Obesity, Metabolic Syndrome and CardioRenal Disease: Evolutionary Considerations. <em>Semin Nephrol</em> <strong>2011</strong>, <em>31</em> (5), 394–399. https://doi.org/10.1016/j.semnephrol.2011.08.002.</li>
  <li>Portis Jr, A. R.; Salvucci, M. E. The Discovery of Rubisco Activase – yet Another Story of Serendipity. <em>Photosynthesis Research</em> <strong>2002</strong>, <em>73</em> (1), 257–264. https://doi.org/10.1023/A:1020423802875.</li>
  <li>Shepon, A.; Eshel, G.; Noor, E.; Milo, R. Energy and Protein Feed-to-Food Conversion Efficiencies in the US and Potential Food Security Gains from Dietary Changes. <em>Environ. Res. Lett.</em> <strong>2016</strong>, <em>11</em> (10), 105002. https://doi.org/10.1088/1748-9326/11/10/105002.</li>
  <li>Naylor, T. W.; Williamson, T.; Trinci, A. P. J.; Robson, G. D.; Wiebe, M. G. Fungal Food. US5980958A, November 9, 1999.</li>
</ol>]]></content><author><name>keith</name></author><category term="industry" /><category term="featured" /><category term="submerged-fermentation" /><category term="bioreactor" /><category term="fungi" /><category term="sticky" /><summary type="html"><![CDATA[What if you could make food out of just basic chemicals and microbes? At a first glance, this seems like a great idea since microrganisms grow quickly, are thermodynamically efficient, and can add value to low-cost chemicals and waste streams.]]></summary><media:thumbnail xmlns:media="http://search.yahoo.com/mrss/" url="https://controlledmold.com/assets/images/quorn/cover.png" /><media:content medium="image" url="https://controlledmold.com/assets/images/quorn/cover.png" xmlns:media="http://search.yahoo.com/mrss/" /></entry><entry><title type="html">Design of the koji room (koji-muro)</title><link href="https://controlledmold.com/design-of-the-koji-room-koji-muro/" rel="alternate" type="text/html" title="Design of the koji room (koji-muro)" /><published>2021-09-07T00:00:00+00:00</published><updated>2021-09-07T00:00:00+00:00</updated><id>https://controlledmold.com/design-of-the-koji-room-koji-muro</id><content type="html" xml:base="https://controlledmold.com/design-of-the-koji-room-koji-muro/"><![CDATA[<p>I hope this post will provide a bit of guidance to anyone trying to build a koji room (also known as the <em>koji muro</em> or <em>sei-kiku shitsu</em>), especially as they scale up to the 50-1500kg/batch range. I’ll go into the details of the history, design, and operation of the koji room.</p>

<p>This is part 2 of “How koji is produced on industrial scales”. (Click here for <a href="https://controlledmold.com/industrial-koji-fermentation-equipment/">Part 1</a>).</p>

<h2 id="introduction">Introduction</h2>

<p>As much as I enjoy learning about 50-ton capacity, fully automated koji making machines, these machines are expensive and fairly out-of-reach for smaller producers looking to enter the ‘micro-brewery’ scale.</p>

<p>That being said, koji rooms (<em>koji muro</em>) are still extremely commonplace. They have a fairly low capital cost, and many of the smaller koji-making machines are based inside of a koji room.</p>

<p>Although it appears to be a fairly simple piece of equipment, there’s a lot of interesting engineering that goes into a koji room. I couldn’t find any English guidance on the topic, but the key concept is essentially HVAC-meets-biology, with numbers and parameters from some key Japanese resources which I’ve cited at the end of this post.</p>

<p>I’ll be delving into the purpose of a koji room, some history, a design example, koji-making machines based within a koji room, and some of the finer details.</p>

<h2 id="design-requirements">Design Requirements</h2>

<p>A properly designed koji room must meet the following design criteria:</p>

<ol>
  <li>Provide a work area that’s easy to sanitize</li>
  <li>Be insulated to prevent condensation</li>
  <li>Maintain a set temperature and humidity</li>
  <li>Be ventilated to remove heat, water, and carbon dioxide from koji growth</li>
</ol>

<p>We’ll take a look at each of these in detail, and how they interact with each other. Firstly, a bit of history.</p>

<h2 id="traditional-koji-rooms-pre-1950s">Traditional Koji Rooms (pre-1950s)</h2>

<p>Breweries prior to the 1950s had virtually zero mechanization, and no modern materials of construction.  During this time, four types of koji rooms were common: basement, half-basement, first floor, or upper floor - based simply on where they were constructed and the flow of material within the brewery.</p>

<p>In basement and half-basement koji rooms, steaming and soaking of the substrate would take place on the ground floor of the brewery and dropped into the koji room below. These rooms were typically lined with plaster, brick, or later concrete, typically in two layers and insulated with a mixture of rice straw and rice husks in-between.</p>

<p>First- and upper-floor koji rooms would also use rice straw and rice husks as insulation, lined between interior wooden (often cedar) walls and exterior walls.</p>

<p><img src="/assets/images/koji_room/Traditionalmuro.png" alt="" width="450" class="center" />
<em>Plan view of a typical traditional first floor koji room (koji-muro) from [1]</em></p>

<p>Without any electric fans, these koji rooms were equipped with one or several pairs of chimneys called <em>ten-mado</em>. By using two chimneys set a different heights, hot moist air from the koji room would escape through the taller chimney and cold outdoor air would enter through the shorter chimney. This design employs a principle called the <strong>stack effect</strong> in HVAC. Operators could adjust a shutter at the bottom of the chimney to control air flow.</p>

<p><img src="/assets/images/koji_room/Tenmado.png" alt="" width="450" class="center" />
<em>Two types of dual chimneys (ten-mado) from [2]</em></p>

<p>There were a few known issues with these older koji rooms. Firstly, they would often be heated by small fires or hot charcoals - and anoxia was a safety issue. Secondly, the rice straw insulation, although effective when dry, loses its thermal insulation properties when wet. They were also magnets for non-desirable organisms including Japan’s brewery enemy No. 1, <em>Bacillus subtilis</em>.</p>

<p>That being said, these koji rooms were effective - and even though they lack most modern HVAC luxuries, the brewery still had access to the two most important pieces of instrumentation: a dry-bulb and wet-bulb thermometer, from which you can calculate all properties of the room air using a <strong>psychrometric chart</strong>.</p>

<h2 id="modern-koji-rooms">Modern Koji Rooms</h2>

<p>Modern koji-rooms solved all the problems with old koji-rooms, and starting in the 1950s, all breweries quickly renovated to employ the following design features, which I’ll quickly summarize:</p>

<ol>
  <li>Electric heaters (often wall-mounted panel heaters or wall-mounted heating coils)</li>
  <li>Modern insulation (fibreglass, mineral wool, PVC foam, styrofoam, or polyurethane foam)</li>
  <li>Modern materials of construction (coated steel, stainless steel)</li>
  <li>Exhaust fans and inlet louvres &amp; dampers</li>
  <li>Humidifiers (often simple household humidifiers or misting nozzles)</li>
</ol>

<p>Despite these upgrades, I’d like to point out that both cedar-lined interior walls and the dual-chimneys (<em>ten-mado</em>) are still commonplace in more traditional breweries - it seems these design elements have withstood the test of time.</p>

<h2 id="design-of-the-koji-room">Design of the Koji Room</h2>

<p>Perhaps the easiest way to explain the design of a koji-room is to go through an example. I don’t want this blog post to become a textbook, so I’ll provide a quick overview of the design criteria first, and present the concepts visually. Anyone looking for calculation details can into to the Appendix at the very end of this post. A basic understanding of <strong>psychrometrics</strong> and <strong>heat transfer</strong> is nice to have for this exercise.</p>

<p>Let’s consider a koji room with the following parameters:</p>

<ol>
  <li>Seishu (<em>sake</em>) rice-koji polished 75% with a capacity of 300kg steamed rice per batch (<em>hiki-komi</em> weight - more on the process <a href="https://controlledmold.com/the-koji-making-process-temperature-mycelium-and-moisture/">here</a>)</li>
  <li>Table koji method (more on methods <a href="https://controlledmold.com/industrial-koji-fermentation-equipment/">here</a>)</li>
  <li>Room conditions are 30°C at 70% Relative Humidity (RH) - constant throughout process</li>
  <li>Outdoor / brewery conditions are 5°C at 50% RH - sake brewing performed in the winter</li>
</ol>

<p>When designing any equipment, we always consider the extreme cases under which it will operate, also known as the <strong>operating window</strong>. For this example, we’re considering operation on a cold day at two times: at the very start of the process immediately after inoculation (when the koji generates no heat) and at the peak of its growth (when the koji generates maximum heat output). This is a bit of an oversimplification, but it will be enough to demonstrate the concept.</p>

<h3 id="design-basis">Design basis</h3>

<p>The table below provides all the key information needed to properly design a koji room (and a koji-making machine) for some well-known substrates:</p>

<p><strong>Table 1 - Design parameters for common koji substrates [4]</strong></p>

<table class="table-clean">
  <thead>
    <tr>
      <th>Substrate</th>
      <th style="text-align: right">Initial Moisture<br /> (%)</th>
      <th style="text-align: right">Initial Specific Volume<br /> (L/kg)</th>
      <th style="text-align: right">Peak Heat Generation <br />(kJ/kg-dry/h)</th>
      <th style="text-align: right">Total Heat Generation<br />  (kJ/kg-dry)</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>Wheat bran &amp; rice husk koji (1:1)</td>
      <td style="text-align: right">50-55</td>
      <td style="text-align: right">9.2</td>
      <td style="text-align: right">158</td>
      <td style="text-align: right">1990-2620</td>
    </tr>
    <tr>
      <td>Wheat bran koji</td>
      <td style="text-align: right">50</td>
      <td style="text-align: right">4.4</td>
      <td style="text-align: right">138</td>
      <td style="text-align: right">1800-2100</td>
    </tr>
    <tr>
      <td>Miso rice koji (polishing ratio 90%)</td>
      <td style="text-align: right">40-45</td>
      <td style="text-align: right">2.5</td>
      <td style="text-align: right">41.8-64.9</td>
      <td style="text-align: right">460-750</td>
    </tr>
    <tr>
      <td>Seishu rice koji (polishing ratio 75%)</td>
      <td style="text-align: right">35</td>
      <td style="text-align: right">2.2</td>
      <td style="text-align: right">30</td>
      <td style="text-align: right">380-440</td>
    </tr>
    <tr>
      <td>Shoyu koji (soybean &amp; wheat 1:1)</td>
      <td style="text-align: right">40-45</td>
      <td style="text-align: right">2.9</td>
      <td style="text-align: right">130</td>
      <td style="text-align: right">1800-2300</td>
    </tr>
  </tbody>
</table>

<div class="callout">
💡 Seishu (sake) rice koji generates the least amount of heat among koji types, for a simple reason: it's the least nutritious. The more you polish rice, the more you strip away lipids, proteins, and minerals which are concentrated in the outer layers. Highly polished rice is little more than a ball of starch with a bit of protein and minerals - so the koji grows slowly. By comparison - shoyu koji, which contains whole soybeans and wheat, is choc full of nutrition and generates over 4x the heat of seishu koji. 
</div>

<p>With these parameters, the remainder of the process is fairly simple: all we have to do is go through four design calculations and determine, in order, the room dimensions, insulation, ventilation, and heating.</p>

<h3 id="1-dimensions">1. Dimensions</h3>

<p>Let’s determine the size of our table (<em>toko</em>) and koji room. To fit 300kg of steamed rice for inoculation (<em>tane-kiri</em>) and at the end of the process (<em>shimai shigoto</em>) at a depth of 5cm, we need a 13.2 square meter table, housed in a koji room measuring about <strong>8m(L) x 5m (W) x 2.2m(H)</strong> in order to fit clearances and other equipment (See Appendix - <a href="#step-1---determine-the-room-and-outdoor-conditions">Step 1</a>, <a href="#step-2---determine-the-koji-amount-dry-weight-and-room-dimensions">Step 2</a>).</p>

<h3 id="2-insulation">2. Insulation</h3>

<p>The koji room’s insulation must be sufficient to keep the room warm AND prevent condensation on the inside surfaces of the room.  If we look at a section of the koji-room wall, overlaid with a steady-state temperature profile, it will look something like this:</p>

<p><img src="/assets/images/koji_room/Wallheattransfer.png" alt="" width="450" class="center" />
<em>Depiction of 1-D steady state heat transfer through a wall</em></p>

<p>Our koji room, sitting at 30°C and 70% RH, will be slightly colder near the inside wall in an area known as the <strong>temperature boundary layer</strong>. The temperature right up against the interior wall <em>must</em> be greater than the dew point of the air in the koji room (at 30°C and 70% RH, the dew point is 23.9°C from a psychrometric chart) or condensation will form.</p>

<p>This is a 1-dimensional heat transfer problem. If we pick 4-inch polyurethane with a total RSI-value of 4.24 (RSI is the metric equivalent of the imperial R-value, a measure of an insulation’s thermal resistance), we can demonstrate that the wall temperature will not fall below 29.3°C. (see Appendix - <a href="#step-3---determine-heat-loss-through-the-room-walls-and-check-for-condensation">Step 3</a>). This might seem like overkill - our design requirement was 23.9°C! But trust me, insulation is cheap relative to the rest of the room; and the heat retention will save costs in the long run.</p>

<p>If the indoor conditions are maintained at 30°C, we can further demonstrate that the heat losses through the wall will total <strong>764W</strong>. To keep the room warm, an equal amount of heat must be supplied. Let’s illustrate the conditions of the koji room at time t=0 after inoculation, during which the koji generates no heat and all openings to the room are closed:</p>

<p><img src="/assets/images/koji_room/01kojiroominitial.png" alt="Heat balance of the koji room immediately after inoculation" width="650" class="center" /></p>

<h3 id="3-ventilation">3. Ventilation</h3>

<p>The koji room’s ventilation must be sufficient to remove excess heat, humidity, and water generated by the koji as it grows. <strong>Removal of water vapour</strong> is usually the limiting design criteria, and requires a greater volume of ventilation than <strong>heat removal</strong> or <strong>carbon dioxide</strong> removal.</p>

<p>Let’s demonstrate which of these three design criteria is our limit, using our example.</p>

<p>First, consider the requirements for <strong>heat removal</strong>. We can show that our 300kg of rice-koji at <em>hiki-komi</em> has a dry weight of 195kg, and will generate 1623W at the peak of its metabolism (see Appendix - <a href="#step-4---determine-the-ventilation-requirements-for-heat-removal">Step 4</a>). Our conductive heat losses were calculated to be 764W, so we know that ventilation for heat removal must equal 859W by exchanging indoor air with cold, outdoor air. This requires a ventilation rate of <strong>41m³/h</strong>.</p>

<p><img src="/assets/images/koji_room/02kojiroomheat.png" alt="Heat balance of the koji room with ventilation sized for heat removal" width="650" class="center" /></p>

<p>Next, let’s consider the requirements for <strong>moisture removal</strong>. About 80% of the heat generated by koji is lost as water vapour, with the remaining 20% lost due to other effects. At the peak of its heat generation, we can show that the our 300kg of koji will emit 2.16kg of water per hour (see Appendix - <a href="#step-5---determine-the-ventilation-requirements-for-moisture-removal">Step 5</a>). By exchanging moist indoor air with dry, outdoor air, we can show that a ventilation rate of <strong>119m³/h</strong> is necessary to remove the moisture emitted by koji.</p>

<p><img src="/assets/images/koji_room/03kojiroomwater.png" alt="Mass balance of the koji room with ventilation sized for moisture removal" width="650" class="center" /></p>

<p>Finally, let’s consider the requirements for <strong>carbon dioxide removal</strong>. Koji has a respiratory coefficient of around 1.0 throughout its entire process, which means that the koji’s peak heat generation of 1623W equates to a CO₂ generation rate of 0.3m³ CO₂/h (see Appendix - <a href="#step-6---determine-the-ventilation-requirements-for-co-removal">Step 6</a>) using the aerobic metabolism equation:</p>

\[C_6H_{12}O_6+6O_2 \rightarrow 6CO_2 +6H_2O+2872kJ/\text{mol} \cdot \text{glucose}\]

<p>If we maintain an indoor carbon dioxide concentration at 4500ppm, which is 500ppm below the OSHA Permissible Exposure Limit (PEL = 5000ppm), and assume an outdoor CO₂ concentration of 500ppm, we can show that the flow rate necessary to remove all generated carbon dioxide is <strong>76m³/h.</strong></p>

<p><img src="/assets/images/koji_room/04kojiroomCO2.png" alt="Mass balance of the koji room with ventilation sized for carbon dioxide removal" width="650" class="center" /></p>

<p>Comparing the three ventilation rates, we can see that <strong>moisture removal</strong>, at 119m³/h, is the limiting design criteria in our scenario.</p>

<h3 id="4-heating">4. Heating</h3>

<p>Knowing that our maximum ventilation rate will be <strong>119m³/h</strong>, let’s size the heating panels of the koji-room. Our conductive heat losses are 764W, and ventilating fresh air at 119m³/h removes 2480W of heat. Our koji generates 1623W of heat at its peak, so our heater must provide at least 1621W of heat (see Appendix - <a href="#step-7---determine-the-heating-requirements">Step 7</a>):</p>

<p><img src="/assets/images/koji_room/05kojiroomfinal.png" alt="Final heat balance the koji room with ventilation with ventilation sized for moisture removal" width="650" class="center" /></p>

<p>And that’s it! At t=0h, our heaters will be running at 764W and our ventilation turned off; and at the peak of growth, our heaters will be running at 1459W and our ventilation at 119m³/h. This is our operating window, and every other point in time will fall somewhere between these two extremes. Achieving steady conditions is the job of whatever <strong>control system</strong> we select. This can range from a basic thermostat and humidistat to a programmable controller.</p>

<h3 id="design-notes">Design Notes</h3>

<p>Engineering is all about precise calculations, then applying a bunch of somewhat arbitrary safety factors, gut intuition, and hoping your control system is tuned well enough. In the example above, I would install an exhaust fan capable of 150m³/h, and heater of 2000W, depending on what’s available from the vendor.</p>

<p>One interesting observation is that, since moisture removal is the limiting factor during the peak of koji growth, we can demonstrate that humidification is not required during koji-making except for the initial stages (by the same logic, if heat removal was the limiting factor, we would not require any heating and would have to supplement humidity).</p>

<p>You might also notice that, if outdoor conditions approach the conditions of the koji room, the ventilation requirements would approach infinity. If it must be done in summer conditions, air conditioning and/or dehumidification is sometimes necessary - but that’s enough calculations for one post.</p>

<p>Ventilation requirements for CO₂ removal are also inversely proportional to the concentration of CO₂ you’d like to maintain in the room. That’s why we want to pick the highest CO₂ concentration safely possible (5000ppm - 500ppm contingency = 4500ppm) to reduce ventilation requirements for CO₂ removal.</p>

<p>Finally, the example above considers 300kg of steamed rice going through the entire growth cycle in one room. It’s common to have two batches going simultaneously in one room, one in the first phase of growth, on the table (<em>toko-momi</em> to <em>mori</em>) and one in the latter stage of growth, on another table; or in trays or boxes placed on shelves (<em>tana</em>) along the perimeter (<em>mori</em> to <em>dekoji</em>). It’s also common to perform the process in multiple, adjacent rooms so they can be controlled separately. These will all affect the final design.</p>

<h2 id="procurement">Procurement</h2>

<p>Great, so how do I actually build a koji room?</p>

<p>The best approach is to procure a <strong>walk-in cooler</strong> with the correct specifications and add the necessary retrofits (exhaust fan, intake damper, heating panels) through an experienced HVAC contractor. Because these are common (and sometimes prefabricated) structures, they are easy to find, relatively inexpensive, and easy to accommodate custom specifications. Since they’re used in food settings, they’ll likely be complaint with local food safety regulations and easy to clean. From what I can tell, many modern koji-rooms in Japan are no more than modified walk-in coolers.</p>

<p>A few general notes when procuring a koji room:</p>

<ol>
  <li>Make sure to follow your local building code, health and safety codes, food safety codes, sanitation practices, fire safety codes, and electrical codes.</li>
  <li>The exhaust duct of any koji room is prone to condensation. If this drips onto the koji, you’ll have a problem. Heating the duct, or installing heat-trace, is recommended.</li>
  <li>Provide room to store cloth, fabrics, and to hang koji-related equipment</li>
  <li>Stainless steel panel heaters are best practice - installed close to the floor of the koji room</li>
  <li>A simple household humidifier or atomizing nozzle is sufficient for humidification</li>
</ol>

<p>To date, I could only find one specialty fabricator of traditional koji rooms; a company called <a href="http://www.bc9.ne.jp/~nittou/about.html">Nitto Kogyosho Co., Ltd.</a> based out of Tochigi, Japan. They still use Akita cedar to build them, with dual chimneys or fans for ventilation.</p>

<h2 id="machines-based-inside-the-koji-room">Machines based inside the koji room</h2>

<p>Many smaller koji-making machines are simply based in a koji room. These machines are simple, and use a blower that sends air intermittently through the bed of koji, controlled by a temperature probe embedded in the koji itself. Every time the koji gets too warm, the blower turns on until the temperature is sufficiently lowered, and repeat. This keeps the koji within a fairly narrow range of temperatures.</p>

<p>However, the blower must operate intermittently - otherwise the entire bed of koji will quickly dry out. More advanced machines like the tent-type machine (<em>ten-maku shiki</em>) recirculate a portion of the air leaving the koji layer to mitigate this drying effect.</p>

<p>Here’s an example of the tent-type koji-making machine:</p>

<p><img src="/assets/images/koji_room/Tentschematic.png" alt="" width="450" class="center" /></p>

<p><img src="/assets/images/koji_room/Tentopen.png" alt="" width="450" class="center" /></p>

<p><img src="/assets/images/koji_room/Tentclosed.png" alt="" width="450" class="center" />
<em>Top to bottom: tent-type koji machine schematic, with the cloth tent removed, and with the cloth tent attached - from Hakuyo Co., Ltd.</em></p>

<p>Here’s a simpler box-type koji-making machine, which does not recirculate any air:</p>

<p><img src="/assets/images/koji_room/Hakkomachine.png" alt="" width="450" class="center" />
<em>A box-type koji-making machine situated in a koji room - from Hakuyo Co., Ltd.</em></p>

<p>There are also unventilated type machines, such as <a href="https://www.fujiwara-jp.com/product/detail.php?id=32&amp;l=en">Fujiwara’s VEX-type machine</a> which simply uses a thin layer of koji and a semi-permeable ePTFE membrane to control moisture removal.</p>

<p>While it may be difficult procure Japanese koji-making equipment overseas, you can look at their product catalogue sheets to get a feel of how they’re designed, their capacities, and the size of the blower.</p>

<p>Some other equipment worth mentioning, primarily used to reduce labour costs:</p>

<h4 id="air-shooter">Air shooter</h4>

<p>A common way to transfer steamed substrate into the koji room, and finished koji out of the room, is to use a device called an <em>air shooter</em>. This is a Japanese <em>wasei-eigo</em> term for what would commonly be called a pneumatic conveying system consisting of: 1) a blower 2) a feed assembly consisting of a hopper, rotary valve, and solids-conveying educator, 3) a bunch of hose, and 4) an outlet cyclone (optional). These are much faster than transferring substrate by hand.</p>

<p><img src="/assets/images/koji_room/Airshooter.png" alt="" width="450" class="center" />
<em>Top: a schematic of an air-shooter (pneumatic conveying system), bottom-left: feed assembly, bottom-right: blower. From [1] and Toyo Shokai Co., Ltd.</em></p>

<h4 id="mixing-kiri-kaeshi-machine">Mixing (kiri-kaeshi) machine</h4>

<p>A quicker way of performing koji mixing is with a mixing (<em>kiri-kaeshi</em>) machine. This is nothing then a solids mixer with a mesh screen for breaking up clumps.</p>

<p><img src="/assets/images/koji_room/Kirikaeshimachine.png" alt="" width="450" class="center" />
<em>A mixing (kiri-kaeshi) machine, from Hakuyo Co., Ltd.</em></p>

<h4 id="load-cells">Load cells</h4>

<p>Some breweries install load cells at the feet of their koji-making table to track substrate weight throughout the process.</p>

<p><img src="/assets/images/koji_room/Loadcell.png" alt="" width="450" class="center" />
<em>A load cell installed underneath a table (toko) in a traditional style room, from Hakuyo Co., Ltd.</em></p>

<h4 id="the-room">The room</h4>

<p>And finally here’s the best picture I could find of a koji room online, which is cedar-lined, uses a modern koji-making machine, and includes wall-mounted panel heaters, a mixing machine, and some other features discussed in this post.</p>

<p><img src="/assets/images/koji_room/Bestroom.jpg" alt="" width="450" class="center" />
<em>A photo of koji room from www.sake-hourai.co.jp: with heating panels (lower side of walls), a room-based koji machine (centre), kiri-kaeshi machine (back of room), thermometer and wet-bulb thermometer (left) and exhaust chimney (roof)</em></p>

<h2 id="conclusions">Conclusions</h2>

<p>There are a lot more details difficult to cover in one post, but hopefully this will give you an idea of how to size, construct, and operate a koji room (<em>koji muro</em>), and even how to build some of the simpler koji-making machines.</p>

<p>If you plan on constructing a koji room, or have constructed one already, I’d love to hear about it! Feel free to reach out to me on twitter <a href="https://twitter.com/Keeifa">@Keeifa</a></p>

<p>This post is part of an ongoing effort to translate Japanese koji-related literature to the west. If you’re interested in this, feel free to reach out to me as well.</p>

<p>In the next post of this series, we’ll look at fully automatic koji making machines.</p>

<h2 id="references">References</h2>

<ol>
  <li><em>Shochu Manufacturing Technology,</em> 3rd ed.; Nishitani, N.; The Brewing Society of Japan:  Tokyo, 2014</li>
  <li><em>Koji Studies</em>, 6th ed.; Hideya, M.; The Brewing Society of Japan: Tokyo, 2018.</li>
  <li>Sato, K. Solid-State Culture (I) Characteristics and the Application of Solid-State Culture. <em>Journal of the Brewing Society of Japan</em> <strong>1992</strong>, <em>87</em> (8), 558–565. https://doi.org/10.6013/jbrewsocjapan1988.87.558.</li>
  <li>Sato, K. Solid-State Culture (II) Controlling of Growth of Microorganism on Solid Media. <em>Journal of the Brewing Society of Japan</em> <strong>1992</strong>, <em>87</em> (12), 874–879. https://doi.org/10.6013/jbrewsocjapan1988.87.874.</li>
  <li>Sato, K. Solid-State Culture (III) Characteristics of Solid-State Fermentors and the Applications. <em>Journal of the Brewing Society of Japan</em> <strong>1993</strong>, <em>88</em> (10), 763–774. https://doi.org/10.6013/jbrewsocjapan1988.88.763</li>
  <li>Nagatani, M. Koji Machinery Equipment Design Method (Part 1). <em>Journal of the Brewing Society of Japan</em> <strong>1979</strong>, <em>74</em> (9), 593–595. https://doi.org/10.6013/jbrewsocjapan1915.74.593.</li>
  <li>Nagatani, M. Koji Machinery Equipment Design Method (Part 2). <em>Journal of the Brewing Society of Japan</em> <strong>1979</strong>, <em>74</em> (10), 653–655. https://doi.org/10.6013/jbrewsocjapan1915.74.653.</li>
  <li>Nagatani, M. Koji Machinery Equipment Design Method (Part 3). <em>Journal of the Brewing Society of Japan</em> <strong>1979</strong>, <em>74</em> (12), 813–815. https://doi.org/10.6013/jbrewsocjapan1915.74.813.</li>
</ol>

<h2 id="appendix">Appendix</h2>

<p>Here are the calculations for the example in this post. This will read like an engineering textbook.</p>

<h4 id="step-1---determine-the-room-and-outdoor-conditions">Step 1 - Determine the room and outdoor conditions</h4>

<p>The design basis will be for a koji room with the following conditions:</p>

<ol>
  <li>Steamed <em>seishu</em> rice, polished 75%, 300kg steamed rice per batch, single room and one batch per room</li>
  <li>Sea level (101.3kPa)</li>
  <li>Indoor conditions - 30°C / 70% RH. From a psychrometric chart, the indoor air has the following conditions:
    <ul>
      <li>Enthalpy: 78.25kJ/kg dry air</li>
      <li>Absolute humidity: 18.8 g water / kg dry air</li>
      <li>Specific volume: 0.885 m³/kg dry air</li>
    </ul>
  </li>
  <li>Outdoor conditions - 5°C / 50% RH - constant throughout the process
    <ul>
      <li>Enthalpy: 11.78kJ/kg dry air</li>
      <li>Absolute humidity: 2.7 g water / kg dry air</li>
      <li>Specific volume: 0.792 m³/kg dry air</li>
    </ul>
  </li>
</ol>

<p>Air flows in this post are using a dry air basis at 30°C / 70% RH (indoor conditions).</p>

<hr />

<h4 id="step-2---determine-the-koji-amount-dry-weight-and-room-dimensions">Step 2 - Determine the koji amount, dry weight, and room dimensions.</h4>

<p>The weight of steamed rice at inoculation (hiki-komi) will be 300kg. If the unsoaked white rice has a moisture of 13%, and the water absorption ratio at inoculation is 34%, the actual water content of the koji is:</p>

\[\%Total\ moisture = \frac{Total\  Water}{Total\  Weight}=\frac{13\%+34\%}{100\%+34\%} = 35.1\%\]

<p>This is in agreement with Table 1. Thus the dry weight of the steamed rice is 194.8kg and its peak heat generation is 5843kJ/h or 1623W (unit conversions not shown).</p>

<p>Let the depth of koji be 5cm. Based on Table 1, 300kg of steamed rice will occupy approximately 660L or 0.66m³ in volume and require a table with an area of 13.2 square meters (geometric calculations not shown). Let’s select a 5.3m x 2.5m table. To allow for clearance and the storage of other items, let’s select a room of 8m (L) x 5m (W). Finally, a ceiling height of 2.0-2.2m is standard - we will select 2.2m (H).</p>

<hr />

<h4 id="step-3---determine-heat-loss-through-the-room-walls-and-check-for-condensation">Step 3 - Determine heat loss through the room walls, and check for condensation</h4>

<p>1-D steady state heat transfer through a wall is given by the following equation:
\(Q=UA(T_1-T_2)\)
Where,</p>

<ul>
  <li>Q = Heat loss (W)</li>
  <li>A = Area (m)</li>
  <li>U = Overall heat transfer coefficient (W/m-K )</li>
  <li>T₁ = Bulk koji room temperature</li>
  <li>T₂ = Bulk outdoor temperature</li>
</ul>

<p>The overall heat transfer coefficient is given by the following equations (the first equation uses thermal conductivity, and the second equation uses R-values, which are often more convenient):</p>

<p>\(\frac{1}{U} = \frac{1}{h_1}+\sum\frac{x_n}{k_n}+\frac{1}{h_2} \\
\frac{1}{U} = \frac{1}{h_1}+\sum RSI_n +\frac{1}{h_2}\)
Where,</p>

<ul>
  <li>$h_1$ = convective heat transfer coefficient (inside room) (W/m-K )</li>
  <li>$h_2$ = convective heat transfer coefficient (outside room) (W/m-K )</li>
  <li>$x_n$ = thickness of insulation layer (m)</li>
  <li>$k_n$ = thermal resistance of insulation layer (W/m-K )</li>
  <li>$RSI_n$ = insulation R value (metric RSI value), equivalent to x/k (m-K/W)</li>
</ul>

<p>Select a 4-inch polyurethane insulation with a total metric RSI value of 4.24 (m-K/W). We’ll use a convective heat transfer coefficient of 8.0 W/m-K on both sides of the wall. Now, we can find the overall heat transfer coefficient:</p>

\[\frac{1}{U} = \frac{1}{8}+4.24 +\frac{1}{8} \text{ (SI Units)}\\
U = 1.3W/m^2\cdot K\]

<p>As well as the heat losses through the wall:</p>

\[Q = (1.3W/m^2\cdot K)(137.2m^2 )(30^\circ C-5^\circ C) \\
Q = 764W\]

<p>Heat flux is simply heat transfer divided by area, or Q/A. In 1-D steady state heat transfer, this term is constant no matter where you are along the wall. 
\(\frac{Q}{A} = h(T_1 - T_w)\)
Where,</p>

<p>$T_w$ is the temperature at the koji room wall</p>

<p>If we solve for the wall temperature:</p>

\[763W = (8W/m^2\cdot K)(137.2m^2)(30^\circ C-T_w)\\
T_w = 29.3^\circ C\]

<p>This is well above the dew point of our koji room, so no condensation is expected.</p>

<hr />

<h4 id="step-4---determine-the-ventilation-requirements-for-heat-removal">Step 4 - Determine the ventilation requirements for heat removal</h4>

<p>This is a heat balance on the koji room:</p>

\[Q_{koji} = Q_{wall} + Q_{vent}\]

<p>Where,</p>

<ul>
  <li>$Q_{koji}$ - peat heat generation from koji (W)</li>
  <li>$Q_{wall}$ - heat losses through the wall (W) (from Step 3)</li>
  <li>$Q_{vent}$ - heat removed via ventilation (W)</li>
</ul>

<p>Heat removed by ventilation is simply flow rate times the difference in the air enthalpy in and out of the system:</p>

\[Q_{vent} = \dot{m}(H_1 -H_2)\]

<p>Where,</p>

<ul>
  <li>ṁ = mass flow rate of air (kg dry air/s)</li>
  <li>H₁ = enthalpy of air leaving the koji room (kJ/kg dry air)</li>
  <li>H₂ = enthalpy of air entering the koji room (kJ/kg dry air)</li>
</ul>

<p>Our heat balance becomes:</p>

\[Q_{koji} = Q_{wall} + \dot{m}(H_1-H_2)\\
1623W = 878W + \dot{m}(78.3kJ/kg - 11.78kJ/kg)\cdot(\frac{1000W}{kJ/s})\\
\dot{m}=0.130 kg\ dry\ air/s\]

<p>We know its specific volume is 0.885 m³/kg dry air so the ventilation rate is 0.0114m³/s or 41.1m³/h (indoor basis).</p>

<hr />

<h4 id="step-5---determine-the-ventilation-requirements-for-moisture-removal">Step 5 - Determine the ventilation requirements for moisture removal</h4>

<p>Mass balance on water:
\(\text{Water generated by koji} = \text{Water removed by ventilation} \\
\frac{FQ_{koji}}{\Delta H_{vap,T_1}}=\dot{m}(W_1-W_2)\)</p>

<p>Where</p>

<ul>
  <li>F = fraction of heat generated by koji removed by evaporative cooling (0.8)</li>
  <li>$ΔH_{vap,T1}$ = heat of vaporization of water at conditions T₁ inside koji room (J/g water)</li>
  <li>W₁ = koji room absolute humidity (g water / kg dry air)</li>
  <li>W₂ = outdoor absolute humidity (g water / kg dry air )</li>
</ul>

<p>Our mass balance becomes:</p>

\[\frac{(0.8)(1623W)}{2429.8J/g} = \dot{m}(18.8\text{g water/kg dry air} - 2.7\text{g water/kg dry air}) \\
m = 0.03317\text{kg dry air/s}\]

<p>We know its specific volume is 0.885 m³/kg dry air so the ventilation rate is 105.7m³/h (indoor basis).</p>

<hr />

<h4 id="step-6---determine-the-ventilation-requirements-for-co-removal">Step 6 - Determine the ventilation requirements for CO₂ removal</h4>

<p>Conversion of aerobic metabolic heat generation to carbon dioxide generation:</p>

\[\begin{align}
&amp; C_6H_{12}O_6+6O_2 \rightarrow 6CO_2 +6H_2O+2872kJ/\text{mol} \cdot \text{glucose} \\
&amp; 2872kJ/\text{mol}\ \text{glucose} \times\frac{1 \text{mol}\ \text{glucose}}{6\text{mol} \ CO_2}= 478.7kJ/\text{mol}\ CO_2 \rightarrow0.00209\text{mol}\ CO_2 /kJ \\
&amp; \frac{0.00209\text{mol}\ CO_2 }{kJ}\times \frac{5843kJ}{h}(Q_{koji})=12.2\text{mol}\ CO_2 /h
\end{align}\]

<p>We can convert to volumetric CO₂ generation using the ideal gas law:</p>

\[\dot{V}_{CO_2,koji}=\frac{\dot{n}RT}{P}\\
\dot{V}_{CO_2,koji} =12.2\text{mol}\ CO_2 /h\times\frac{(8.314J\text{mol}^{-1}K^{-1})(303.15K)}{101300Pa} \\
\dot{V}_{CO_2,koji} = 0.304m^3/h\]

<p>Mass balance on carbon dioxide:</p>

\[CO_2 \text{ generated by koji} = CO_2 \text{ removed by ventilation} \\
\dot{V}_{CO_2,koji} =\dot{V}(C_1-C_2)\]

<p>Where,</p>

<ul>
  <li>$m_{CO₂,koji}$  = CO₂ generated by koji (m³/h)</li>
  <li>V(dot) = ventilation flow rate (m³/h)</li>
  <li>C₁ = concentration of CO₂ inside koji room (fraction)</li>
  <li>C₂ = concentration of CO₂ outside (fraction)</li>
</ul>

<p>Our mass balance becomes:</p>

\[0.304m^3/h = \dot{V}(0.004500-0.000500) \\
\dot{V}=75.9m^3/h\]

<hr />

<h4 id="step-7---determine-the-heating-requirements">Step 7 - Determine the heating requirements</h4>

<p>Heat balance on the koji room:</p>

\[Q_{koji}+Q_{heater}=Q_{wall}+Q_{vent}\]

<p>Where $Q_{heater}$ = heat required by the heater</p>

<p>Our heat balance becomes:</p>

\[1623W + Q_{heater} =763W + (0.03317kg\ \text{dry air}/s)(78.3kJ/kg - 11.78kJ/kg)\cdot(\frac{1000W}{kJ/s}) \\
Q_{heater} = 1621W\]

<hr />]]></content><author><name>keith</name></author><category term="koji" /><category term="koji" /><category term="solid-state-fermentation" /><category term="scale-up" /><summary type="html"><![CDATA[I hope this post will provide a bit of guidance to anyone trying to build a koji room (also known as the koji muro or sei-kiku shitsu), especially as they scale up to the 50-1500kg/batch range. I’ll go into the details of the history, design, and operation of the koji room.]]></summary><media:thumbnail xmlns:media="http://search.yahoo.com/mrss/" url="https://controlledmold.com/assets/images/koji_room/kojiroomthumb.png" /><media:content medium="image" url="https://controlledmold.com/assets/images/koji_room/kojiroomthumb.png" xmlns:media="http://search.yahoo.com/mrss/" /></entry><entry><title type="html">Flavourzyme: a purified enzyme mixture from A. oryzae</title><link href="https://controlledmold.com/flavourzyme-a-purified-enzyme-mixture-from-a-oryzae/" rel="alternate" type="text/html" title="Flavourzyme: a purified enzyme mixture from A. oryzae" /><published>2021-06-04T00:00:00+00:00</published><updated>2021-06-04T00:00:00+00:00</updated><id>https://controlledmold.com/flavourzyme-a-purified-enzyme-mixture-from-a-oryzae</id><content type="html" xml:base="https://controlledmold.com/flavourzyme-a-purified-enzyme-mixture-from-a-oryzae/"><![CDATA[<p>Flavourzyme™️ is a trademark name given to a mixture of refined proteolytic enzymes branded by Danish biotech company Novozymes A/S. It’s produced by the submerged fermentation of our friend <em>A. oryzae</em> - in this post we’ll look at what it is, and how it’s made.</p>

<h3 id="what-is-flavourzyme">What is Flavourzyme?</h3>

<p>The modern food processing industry uses proteolytic enzymes everywhere: from accelerating cheese aging, modifying breads, as a flavouring (any time you see “hydrolysed vegetable protein”), processed meats, in detergents… the list goes on and on.</p>

<p>Flavourzyme is a specific mixture of proteolytic enzymes extracted from an <em>Aspergillus oryzae</em> liquid culture. What sets it apart from other industrial proteolytic enzymes is that it contains a cocktail of both endo- and exo-peptidases that can efficiently convert proteins into something with more umami, making it useful for applications requiring the development of flavour [1].</p>

<p>Why does this matter? If a protein is fully decomposed into its constituent amino acids, it tends to taste savoury in the presence of salt - as long as there are sufficient glutamates relative to other amino acids. However, if a protein is only partially decomposed, specific peptides can be extremely bitter: for example, the free amino acids Leucine (Leu) and Phenylalanine (Phe) are slightly bitter (with a tasting threshold of 15-20mM), but their simple dipeptide forms, Leu-Leu and Leu-Phe, are <strong>ten times</strong> more bitter [2].</p>

<p>This is why Flavourzyme works well in food applications requiring umami: it contains endopeptidases, which snip peptide bonds in the <strong>middle</strong> of a peptide chain, as well as exopeptidases, which snip peptide bonds at the <strong>terminal ends</strong> of a peptide chain.</p>

<p><img src="/assets/images/flavourzyme/Endo-Exo-Peptidases.png" alt="how endo and exo peptidases act on proteins" width="550" class="center" />
<em>Simplified diagram of the action of endo- and exo-peptidases.  Exopeptidases can cut at the terminal peptide bond, one or two bonds away from the terminal peptide bond, or right in the middle of a dipeptide.</em></p>

<p>Working together in the correct proportion of endopeptidases and exopeptidases, you end up with an enzyme cocktail that yields plenty of amino acids without accumulating any short, bitter peptides.</p>

<div class="callout">
💡 A quick sidenote on proteolytic enzymes: first of all, endopeptidases are typically very specific - they will cut only a specific bond at a specific pH (for example, Trypsin cuts ONLY after Arg OR Lys UNLESS followed by Pro). Secondly, every time an endopeptidase cuts, it frees up two new terminal ends for exopeptidases to cut. Exopeptidases that cut on the amino end are called aminopeptidases, and those that cut on the carboxy end are called carboxypeptidase.
</div>

<p>But wait - how might you make Flavourzyme? And how on earth do you grow <em>A. oryzae</em> in a liquid?</p>

<h3 id="growing-a-oryzae-in-a-liquid---upstream-processing">Growing <em>A. oryzae</em> in a liquid - upstream processing</h3>

<p>In bioprocess engineering, the process of cultivating a culture is called upstream processing (refining the resultant products is called downstream processing, which we’ll discuss later in this article). For <em>A. oryzae</em>, the process can be a little more involved than growing unicellular organisms such as yeast or bacteria.</p>

<p>Information here is based off a paper from researchers at the University of Denmark, working at the Novozymes A/S Fermentation Pilot Plant, who wanted to develop a mathematical model for enzyme production with <em>A. oryzae</em> in submerged culture [3]. Even though they scrubbed some critical units and sig-figs from their research (for proprietary reasons), it was still enough to piece together details of the process. The process is similar to any other submerged, septic, aerated culture:</p>

<p><img src="/assets/images/flavourzyme/Bioreactor.png" alt="Process diagram of an aerated bioreactor" width="450" class="center" />
<em>Process diagram of an aerated bioreactor</em></p>

<p>A seed train is propagated from a pure culture of A. oryzae and then added to an aerated, stirred bioreactor and allowed to grow. The substrate is typically a mixture of food industry by-products such as starch, wheat bran, corn steep powder, and minerals. Once the culture reaches a certain point, the bioreactor is harvested to yield a crude mixture of enzymes, mycelia, and leftover substrate.</p>

<p>There’s a few quirks to this specific culture method, which applies to some other filamentous fungi fermentations:</p>

<h4 id="1-use-a-fed-batch-configuration">1. Use a fed-batch configuration</h4>

<p>The bioreactor is primed with substrate and inoculum around 70% full, then additional substrate is slowly dosed in. This is done so that substrate is consumed at basically the same rate that it’s fed into the bioreactor, which keeps glucose in the environment low. As a result, enzyme yields are high by avoiding <em>carbon catabolite repression</em> (which you can read about <a href="https://controlledmold.com/the-koji-making-process-temperature-mycelium-and-moisture/">here</a> and plenty of oxygen is available in the medium to prevent anaerobic metabolism.</p>

<h4 id="2-filamentous-fungi-make-everything-viscous">2. Filamentous fungi make everything viscous</h4>

<p>Filamentous fungi don’t normally grow in a submerged environment. The mycelia will aggregate to form pellets or clumps - which reduces substrate uptake and enzyme secretion. To complicate things, a bioreactor full of filamentous mycelia will make the mixture more and more viscous. This makes stirring harder, and also reduces the ability of air bubbles to mix and for oxygen to dissolve. By necessity, oxygen transfer drops as the mycelia grows.</p>

<h4 id="3-use-a-pulsed-paused-feeding-mode">3. Use a pulsed-paused feeding mode</h4>

<p>In addition to using a fed-batch setup, researchers have also found that pulsed-paused feeding, in which feeding of the substrate is turned on/off intermittently in 300s cycles, improves the fungi’s morphology. The rationale is that under starvation considerations, fungal mycelium are known to autolyse (self-digest) older parts of the mycelium network to re-allocate resources to the growing tip. While this mechanism is yet to be studied in detail, it’s been shown that you get better morphology and lower viscosity with pulsed-paused feeding, without sacrificing enzyme productivity [4].</p>

<p>Once the upstream process is finished, you end up with a crude liquid containing fungi, enzymes, spent substrate, and all sorts of metabolites.</p>

<h3 id="refining-the-enzymes---downstream-processing">Refining the enzymes - downstream processing</h3>

<p>In <em>downstream processing</em>, we want to refine out the enzymes only. While I couldn’t find any literature specifically on the downstream processing of flavourzyme, reading textbook on enzymes was enough to cover the basics [5]. Then, a producer of a Flavourzyme-like product on Alibaba just gave me their HACCP flow diagram, so that made things easy. Here’s one example of how it can be done:</p>

<h4 id="1-filter-press">1. Filter Press</h4>

<p>The first step is to filter out the mycelia, cell debris, any any other large solids. A <em>filter press</em> is used here with perlite, followed by diatomaceous earth, as a filtering aid. The solid matter is captured in the filter aid and removed as filter cake - leaving behind an enzyme mixture.</p>

<p><img src="/assets/images/flavourzyme/Deadendfiltration.png" alt="Removing large solids from the raw liquid using perlite and diatomaceous earth " />
<em>Schematic of dead-end filtration with a filter aid - elements not to scale</em></p>

<p>Note that filtration is the first step because we are dealing with <strong>extracellular</strong> enzymes. If the product was an <strong>intracellular</strong> enzyme, this step would be preceded by a ‘disruption’ step to break apart cells.</p>

<h4 id="2-ultrafiltration-concentration">2. Ultrafiltration Concentration</h4>

<p>At this point, our enzyme mixture is still at a fairly low concentration. Another type of filtration called <em>ultrafiltration</em> is used to remove water. By passing the liquid across a filter of the appropriate size (in this case, a filter rated around 10kDa), water will permeate through, leaving behind a more concentrated enzyme solution. This can be repeated until the desired concentration is achieved.</p>

<p><img src="/assets/images/flavourzyme/Ultrafiltration.png" alt="Removing water from the resulting filtrate" />
<em>Schematic of cross-flow ultrafiltration - elements not to scale</em></p>

<h4 id="3-final-steps">3. Final Steps</h4>

<p>I won’t go into too much detail here, but the final mixture can either be dried by spray-drying with the help of a carrier solid (typically sodium chloride), or packaged as a liquid after passing through a sterilizing filter and adding stabilizers and preservatives.</p>

<h3 id="the-final-product">The Final Product</h3>

<p>If everything’s gone well so far we should have a powerful enzyme mixture rated for 500 LAPU/g (one LAPU is the amount of enzyme which hydrolyzes 1 µmol of L-leucine-pnitroanilide per minute - this is one of many enzyme assays for proteolytic enzymes) and 6% w/w of pure enzyme protein, depending on formulation [6]. By comparison, unrefined shoyu koji contains around 0.2% w/w protease [7].</p>

<p>The specific mixture of enzymes identified by researchers is as follows [1]:</p>

<ul>
  <li>leucine aminopeptidase A</li>
  <li>leucine aminopeptidase 2</li>
  <li>dipeptidyl peptidase 4</li>
  <li>dipeptidyl peptidase 5</li>
  <li>neutral protease 1</li>
  <li>neutral protease 2</li>
  <li>alkaline protease 1</li>
  <li>α-amylase A type 3</li>
</ul>

<p>Interestingly, α-amylase ends up present as a by-product - I suppose it’s difficult to get <em>A. oryzae</em> not to produce any amylases.</p>

<h3 id="applications">Applications</h3>

<p>Novozymes A/S won’t sell to non-commercial or non-institutional end-users, so I had to import an imitation product from Alibaba. Here’s what it looks like:</p>

<p><img src="/assets/images/flavourzyme/flavourzyme_photo.jpg" alt="A Flavourzyme-like product from Alibaba" width="550" class="center" />
<em>A Flavourzyme-like product from Alibaba</em></p>

<div class="callout">
⚠ Before we continue, a quick note on safety. Being a pure enzyme and protein product, Flavourzyme can irritate the skin, eyes, and mucous membranes. When working with this product, make sure to wear respiratory protection, eye protection, and clean spills with water (without splashing) or by vacuum with a HEPA filter. Read the MSDS!
</div>

<p>I plan to use this to make ultra-fast garums - which has been attempted by the guys at Noma, but using <a href="https://www.instagram.com/p/Bsd-8kVhQdz/?hl=en">pork pancreatic enzymes</a>. Also, credit goes to Preston at <a href="https://www.culinarycrush.biz/all/i-bought-industrial-protease-enzymes">Culinary Crush</a> for inspiring this post. A few more applications I will test:</p>

<ul>
  <li>Accelerated miso / shoyu</li>
  <li>High-fat miso (flavourzyme contains no lipases)</li>
  <li>Accelerated cured tofu</li>
  <li>Home-made Maggi sauce (Red-cap maggi sauce sold in Canada is essentaily wheat gluten + enzymes and additives)</li>
  <li>Hydrolyzing plant-based milks</li>
</ul>

<p>As a quick side note, Japanese brewers today don’t add pure enzyme mixtures to misos or shoyus - at least not for more expensive types. The reason is unclear, but it has been tested in the past [8].</p>

<p>The other question is, of course, can I brew this at home?</p>

<h3 id="references">References</h3>

<ol>
  <li>
    <p>Merz, M.; Eisele, T.; Berends, P.; Appel, D.; Rabe, S.; Blank, I.; Stressler, T.; Fischer, L. Flavourzyme, an Enzyme Preparation with Industrial Relevance: Automated Nine-Step Purification and Partial Characterization of Eight Enzymes. <em>Journal of Agricultural and Food Chemistry</em> <strong>2015</strong>, <em>63</em>. https://doi.org/10.1021/acs.jafc.5b01665.</p>
  </li>
  <li>
    <p>Maehashi, K.; Huang, L. Bitter Peptides and Bitter Taste Receptors. <em>Cell Mol Life Sci</em> <strong>2009</strong>, <em>66</em> (10), 1661–1671. https://doi.org/10.1007/s00018-009-8755-9.</p>
  </li>
  <li>
    <p>Albaek, M. O.; Gernaey, K. V.; Hansen, M. S.; Stocks, S. M. Modeling Enzyme Production with Aspergillus Oryzae in Pilot Scale Vessels with Different Agitation, Aeration, and Agitator Types. <em>Biotechnol Bioeng</em> <strong>2011</strong>, <em>108</em> (8), 1828–1840. https://doi.org/10.1002/bit.23121.</p>
  </li>
  <li>
    <p>Bhargava, S.; Wenger, K. S.; Marten, M. R. Pulsed Feeding during Fed-Batch Aspergillus Oryzae Fermentation Leads to Improved Oxygen Mass Transfer. <em>Biotechnol Prog</em> <strong>2003</strong>, <em>19</em> (3), 1091–1094. https://doi.org/10.1021/bp025694p.</p>
  </li>
  <li>
    <p><em>Enzymes in Industry: Production and Applications</em>, 3rd ed.; Aehle, W., Ed.; Wiley-VCH: Weinheim, 2007.</p>
  </li>
  <li>
    <p>Flavourzyme 500MG Product Data Sheet - Novozymes A/S. Dated 2017-08-24. Accessed 2021-06-06 from: https://www.ibric.org/myboard/view.php?Board=scicafe000352&amp;filename=0003423_1.pdf&amp;id=3423&amp;fidx=1.</p>
  </li>
  <li>
    <p><em>Koji Studies</em>, 6th ed.; Hideya, M.; The Brewing Society of Japan: Tokyo, 2018.</p>
  </li>
  <li>
    <p>Mochizuki, T. About the use of enzymes in miso brewing. <em>Journal of the Brewing Society of Japan</em> <strong>1969</strong>, <em>64</em> (5), 423–430. https://doi.org/10.6013/jbrewsocjapan1915.64.423.</p>
  </li>
</ol>]]></content><author><name>keith</name></author><category term="food" /><category term="enzymes" /><category term="protease" /><category term="food" /><summary type="html"><![CDATA[Flavourzyme™️ is a trademark name given to a mixture of refined proteolytic enzymes branded by Danish biotech company Novozymes A/S. It’s produced by the submerged fermentation of our friend A. oryzae - in this post we’ll look at what it is, and how it’s made.]]></summary><media:thumbnail xmlns:media="http://search.yahoo.com/mrss/" url="https://controlledmold.com/assets/images/flavourzyme/Cover.png" /><media:content medium="image" url="https://controlledmold.com/assets/images/flavourzyme/Cover.png" xmlns:media="http://search.yahoo.com/mrss/" /></entry><entry><title type="html">A bioreactor for everyone</title><link href="https://controlledmold.com/a-bioreactor-for-everyone/" rel="alternate" type="text/html" title="A bioreactor for everyone" /><published>2021-03-08T00:00:00+00:00</published><updated>2021-03-08T00:00:00+00:00</updated><id>https://controlledmold.com/a-bioreactor-for-everyone</id><content type="html" xml:base="https://controlledmold.com/a-bioreactor-for-everyone/"><![CDATA[<p>When I first got started in fermentation, I was a bit lost. There’s no “how to learn fermentation 101” Medium articles, unfortunately. I think most people learn via their academic background in biology, and fermentation is just an applied subtopic in biology. Hence most discussions about fermentation online are experts talking to other experts. These discussions are often feel like you need a full labratory to practice their advice:  -80℃ freezers, costly reagents, pre-sterilized tools, etc. - not to mention the skills required to handle these pieces of equipment. This has put, in my opinon, a large gap between biology hobbyists/new entrants and professionals.</p>

<p>While one foot is in biology, my other foot is in computer programming. Unlike biology, the gap between professional and hobbyist in computer programming  is near nil. A hobbyist has all the tools available to a professional: access to cloud computing, software development environments, free and open source software, etc. Even cost is not an issue: a full computer can be bought for as little as $15. This near-universal access to programming has a fly-wheel effect: the more people programming ⇒ the more tools available ⇒ the more accessible programming becomes ⇒ the more people programming, and so on. It’s remarkable to think about it, but modern computers are only ~40 years old. We are still in the infancy of computing.</p>

<p>So why is “hobbyist” biology lagging, even though the field is much older? This is a question I’d like to give more thought before answering, but certainly the limited access to tools and training is near the top of the list. This was a motivating reason to build the <a href="https://pioreactor.com/">Pioreactor</a> - a bioreactor that is acccessible by design. On this blog, we gone through some of the beginnings of the Pioreactor (see <a href="https://controlledmold.com/categories.html#bioreactor-project">here</a>), and we’ve decided to double down on this idea and polish it further. The end result is a remarkably <strong>fun</strong> tool, and we are still discovering new ways to use it. For example, I’ve been playing with the idea of directed evolution to design new microbial behaviours. We’ve also developed of a novel way to track the progress of home lactofermentation using a Pioreactor (blog post coming soon).</p>

<p>If you are interested in learning more about the Pioreactor, sign up for updates <a href="https://pioreactor.com">here</a>!</p>]]></content><author><name>cam</name></author><category term="bioreactor-project" /><category term="IoT" /><category term="bioreactor" /><category term="sensors" /><summary type="html"><![CDATA[When I first got started in fermentation, I was a bit lost. There’s no “how to learn fermentation 101” Medium articles, unfortunately. I think most people learn via their academic background in biology, and fermentation is just an applied subtopic in biology. Hence most discussions about fermentation online are experts talking to other experts. These discussions are often feel like you need a full labratory to practice their advice: -80℃ freezers, costly reagents, pre-sterilized tools, etc. - not to mention the skills required to handle these pieces of equipment. This has put, in my opinon, a large gap between biology hobbyists/new entrants and professionals.]]></summary><media:thumbnail xmlns:media="http://search.yahoo.com/mrss/" url="https://controlledmold.com/assets/images/bioreactor_project/pioreactor/Group_6_copy.png" /><media:content medium="image" url="https://controlledmold.com/assets/images/bioreactor_project/pioreactor/Group_6_copy.png" xmlns:media="http://search.yahoo.com/mrss/" /></entry><entry><title type="html">The koji making process: temperature, mycelium, and moisture</title><link href="https://controlledmold.com/the-koji-making-process-temperature-mycelium-and-moisture/" rel="alternate" type="text/html" title="The koji making process: temperature, mycelium, and moisture" /><published>2020-12-16T00:00:00+00:00</published><updated>2020-12-16T00:00:00+00:00</updated><id>https://controlledmold.com/the-koji-making-process-temperature-mycelium-and-moisture</id><content type="html" xml:base="https://controlledmold.com/the-koji-making-process-temperature-mycelium-and-moisture/"><![CDATA[<p>Koji generates a considerable amount of heat as it grows - if you’ve ever made koji at home, it’s almost unsettling to feel the warmth of a koji cake. Controlling that heat, along with water content, mixing, and ventilation, is crucial to produce a quality product. In this article, we’ll dive into how temperature and other factors are both controlled and exploited in the koji-making process.</p>

<h3 id="the-temperature-curve">The Temperature Curve</h3>

<p>This is an example of a temperature - time curve for sake koji. There are also different curves for mirin, miso, shochu, etc., and for different manufacturing methods, but the key concepts apply for all applications.</p>

<p><img src="/assets/images/koji_temp_curve/Graph 1.PNG" alt="graph of temperature" width="650" class="center" /></p>

<p>Note that this chart refers to the substrate temperature, measured inside the bed of koji, and not the room itself. But why is there such a complicated temperature progression? Can’t we just set a single temperature and walk away?  To explain this temperature curve, there are a few things that need to be kept in mind:</p>

<ol>
  <li>As the koji grows, it will generate heat - approximately <strong>100 kcal/kg</strong> for rice koji over its life cycle and <strong>7kcal/kg/hr</strong> at its peak (dry weight basis) - and even more for other types of koji</li>
  <li>The optimal temperature for spore <strong>germination</strong> and <strong>growth</strong> are different</li>
  <li>The optimal temperature for <strong>protease</strong> and <strong>amylase</strong> formation are different</li>
</ol>

<p>I’ve already described the effects of 2. and 3. in a <a href="https://controlledmold.com/temperature-water-content-and-polishing-effects-on-koji/">previous post</a>, but in short: mycelial growth is optimal at 35°C, amylase is favoured between 35°C-40°C, and protease is favoured below 35°C. As for 1., a combination of mixing, ventilation, and adjustments to the bioreactor controls (for machine-made koji) removes heat.</p>

<p>Let’s annotate the temperature-time curve and add the names of the traditional processes to this chart:</p>

<p><img src="/assets/images/koji_temp_curve/Graph 2.PNG" alt="graph of temperature" width="650" class="center" /></p>

<p>The entire process can roughly be broken down into three periods:</p>

<ol>
  <li>For the first 18-22h, the rice is wrapped up and piled into a mound after inoculation <em>(toko-momi)</em>. This is done to keep the pile of rice warm and to retain as much moisture as possible. The koji spores germinate within 6-8 hours, then the pile of koji begins to warm up rapidly as the koji generates its own heat. In the middle of this period at the 10-12h mark, the koji is usually mixed once <em>(kiri-kaeshi)</em> to redistribute moisture between the inside and outside of the pile.</li>
  <li>A step called <em>mori</em> is performed after the initial 18-22h. The pile of rice is distributed into trays, boxes, or spread thinly on another table - depending on the koji-making method. The koji begins to generate a lot of heat at this point, so spreading it out helps dissipate heat and provide air. Two more mixing steps are usually performed at this stage <em>(naka-shigoto</em> and <em>shimai-shigoto)</em>.</li>
  <li>In the final phase of growth, the koji is well-established and enzyme production is the most active. At this stage, the temperature must be maintained according to the type of enzymes needed. Once complete, the act of removing the koji from the room is called <em>dekoji</em> and it is subsequently cooled down</li>
</ol>

<p>A few caveats here:</p>

<p>The graph above is for sake koji only, which finishes at a high temperature to produce a lot of amylase and suppress proteases. I’ve compiled all the temperature curves I have <a href="https://imgur.com/a/q94PGBw">here</a>. Sometimes there will be only two distinct temperature phases if the temperature requirements of periods 2. and 3. are identical.</p>

<p>The exact shape and slope of the curve also depends on the koji-making method (more on methods <a href="https://controlledmold.com/industrial-koji-fermentation-equipment/">here</a>). Machine-made koji will have very accurate temperature control, while traditional and manual methods will show a lot more variability and rely on frequent mixing to bring the temperature down.</p>

<p>And lastly, temperature is just one part of the picture. As the koji progresses, two things need to happen:</p>

<ol>
  <li>
    <p>The koji mycelium needs to grow on the <strong>surface</strong> and <strong>interior</strong> of each grain</p>
  </li>
  <li>
    <p>The koji needs to have a proper amount of <strong>moisture</strong></p>
  </li>
</ol>

<h3 id="mycelia-and-haze">Mycelia and “Haze”</h3>

<p>Japanese brewers have a term called <em>“haze”</em> (pronounced like <em>saké</em>) that describes the state of mycelial growth observed on each grain of koji. The term is further distinguished into two concepts:</p>

<ol>
  <li><em>Haze-mawari</em> - how to koji mycelium spreads on a grain’s <strong>surface</strong></li>
  <li><em>Haze-komi</em> - how the koji mycelium penetrates into the grain’s <strong>interior</strong></li>
</ol>

<p>Koji with a high enzyme concentration must have good growth on the surface of the grain, within the centre of the grain, and have the proper moisture content. Brewers have different terms to describe the final product koji’s visible surface and interior growth:</p>

<table class="table-clean">
  <thead>
    <tr>
      <th>Visual Indicators</th>
      <th>Low interior growth</th>
      <th>Good interior growth</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td><strong>Low surface growth</strong></td>
      <td><em>Haze-ochi</em> (“missed haze”)</td>
      <td><em>Tsuki-haze</em> (“speckled haze”)</td>
    </tr>
    <tr>
      <td><strong>Good surface growth</strong></td>
      <td><em>Nuri-haze</em> (“painted haze”)</td>
      <td><em>Sou-haze</em> (“total haze”)<br /><em>Baka-haze</em> (“fool’s haze”)</td>
    </tr>
  </tbody>
</table>

<p>For most applications, we are aiming to produce <em>sou-haze</em>, which has good surface growth and interior growth to maximize enzyme production. <em>Tsuki-haze</em> is also used in applications requiring good fragrance (such as <em>saké</em>).</p>

<p>It is also possible to achieve good surface growth and interior growth - but still end up with subpar koji. This is called <em>baka-haze</em> or “fool’s haze” - in which mycelial growth is visually good but enzyme yields are comparatively low. This is because the koji’s water content is too high, leading the inhibition of enzyme production via the <strong>glucose effect</strong>.</p>

<p>So, in order to achieve good mycelial growth and good enzyme yield, moisture needs to be controlled accurately.</p>

<h3 id="water-the-weight-ratio-and-the-glucose-effect">Water, the Weight Ratio, and the Glucose Effect</h3>

<p>Water has several important functions, and interacts with both temperature, growth, and enzyme production.</p>

<p>Up to a practical limit (i.e. never boil your rice) growth rate is always proportional to water content, and high water content encourages the germination of koji spores.</p>

<p>However, a high water content also inhibits enzyme production. The reason for this is two-fold. Firstly, there is a mechanism called the <strong>glucose effect</strong> (which is a misnomer, see below). In the presence of abundant glucose, the genes encoding for amylolytic enzymes are switched off. This is a smart adaptation by the microorganism: if there’s already a lot of glucose, why make more enzymes? Secondly, higher levels of moisture increase the activity of enzymes. This is a common effect across all enzymes.</p>

<p>So, in the presence of plenty of water, enzyme activity increases, and the koji’s enzymes end up liberating lots of glucose from the starchy substrate. The glucose is transported back to the koji mycelium, and this triggers the inhibition effect.</p>

<div class="callout">

💡 The formal name for this process is carbon catabolite repression, which is common to many micro-organisms, not just koji, and the mechanism can occur with other carbon sources, not just glucose. There is also an opposite regulatory mechanism: the koji will produce more amylolytic enzymes in the presence of maltose and oligo-saccharides. I'll take a detailed look at these and other mechanisms in the future. Also, using some genetic engineering - some of these mechanisms can be deleted or altered entirely.

</div>

<p>Finally, recall that rice koji can generate up to <strong>7kcal/kg/hr</strong> during the peak of its growth - meaning that without heat removal, the koji’s temperature can increase by roughly 7°C in a single hour and the whole batch would be cooked in under 2 hours. Thankfully, with adequate ventilation heat can be removed by evaporative cooling (the latent heat of vaporization of water is <strong>578kcal/kg</strong> at 34°C). In a single hour, the heat generated by 1kg of koji can be removed by allowing roughly 12g of water to evaporate.</p>

<p>Of course, some of the heat is removed by other mechanisms, like convection and conduction, but it can be shown that the majority of heat removal comes from evaporative cooling. It can also be demonstrated that evaporative cooling is the limiting factor when it comes to air supply for koji growth - more so than oxygen requirements or carbon dioxide removal. More on this in another post.</p>

<p>To keep track of the koji’s moisture content, brewers simply weigh the steamed rice or koji, at whatever stage it’s at, and calculate its weight ratio:</p>

\[\text{Weight ratio (%)}= \frac{ \text{Substrate weight} -  \text{Dry rice weight}}{\text{Dry rice weight}} \times 100\%\]

<p>Note that this is not exactly the same as total water content, since white rice already contains 10-15% moisture, and the koji’s metabolic processes also produce water. In reference to soaked and steamed substrate, it is also referred to as the water absorption rate or ratio.</p>

<p>The weight ratio is usually measured at these key steps. The values for sake-koji on short-grain rice are provided here - for other applications, see the end of this post.</p>

<ol>
  <li>After soaking: 28%</li>
  <li>After steaming: 38%</li>
  <li>At inoculation: 33%</li>
  <li>Final product: 17-18%</li>
</ol>

<p>Now let’s add a weight ratio progression to the temperature curve:</p>

<p><img src="/assets/images/koji_temp_curve/Graph 3.PNG" alt="graph of temperature" width="650" class="center" /></p>

<p>Here’s how the water content progresses during each period of growth:</p>

<ol>
  <li>During germination and the initial growth period, the wrapped-up mound of koji retains a lot of moisture and allows the koji to germinate and grow rapidly. Only a small amount of moisture is lost.</li>
  <li>As the koji’s growth takes off, the heat generated by the koji’s metabolism is removed by evaporative cooling and its weight begins to decline rapidly. Koji producers will slightly reduce the room’s relative humidity and/or increase ventilation to promote cooling.</li>
  <li>During the last phase of growth, the water content is sufficiently low that the koji will produce plenty of enzymes without too much glucose effect. Of course, it cannot be so low that the koji dries out and stop growing entirely.</li>
</ol>

<p>Chapter IV of Koji Studies sums it up nicely:</p>

<p>It can be said that the mystery of making koji is “to harmonize the drying rate of steamed rice with the growth rate of koji.”</p>

<h3 id="notes--conclusions">Notes &amp; Conclusions</h3>

<h4 id="how-much-of-this-matters-commercial-koji-vs-homemade-koji">How much of this matters? Commercial Koji vs Homemade Koji</h4>

<p>For home koji makers: just focus on growing koji before worrying about optimal enzyme yields. However, the reference weight ratios and temperatures can be used to help troubleshoot the process.</p>

<p>One of the reasons commercial producers are so meticulous about these parameters is because it affects their product quality and bottom line. Koji takes more time, labour, and energy compared to the application’s non-koji ingredients, so a good enzyme yield is key. For those of us making koji at home, if enzyme yield is low, you can compensate by just using a comparatively high amount of koji. If koji grows too slowly, we can just wait longer.</p>

<p>I also suspect that because commercial producers are dealing with larger amounts of koji overall, their production is less susceptible to moisture loss, and more susceptible to huge swings in temperature (more koji = less surface area per unit volume).</p>

<h4 id="reference-temperature-curves">Reference Temperature Curves</h4>

<p>I’ve posted all the temperature curves accumulated to date here:</p>

<p><a href="https://imgur.com/a/q94PGBw">Koji-making temperature curves</a></p>

<p>It includes curves for sake, miso, mirin, shoyu, and shochu. Again, shoyu and shochu work slightly differently than described in this post but I’ll look at them in detail another time.</p>

<h4 id="reference-weight-ratios">Reference Weight Ratios</h4>

<p>Here is general table outlining the temperatures and weight ratios for different applications.</p>

<table class="table-clean">
  <thead>
    <tr>
      <th>Application</th>
      <th>Substrate - Polishing Ratio</th>
      <th>After soaking</th>
      <th>After steaming</th>
      <th>Completion (Dekoji)</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>Sake</td>
      <td>Rice - 50-75%</td>
      <td>28-30%</td>
      <td>38-40%</td>
      <td>17-19%</td>
    </tr>
    <tr>
      <td>Mirin</td>
      <td>Rice - 83-90%</td>
      <td>28-30%</td>
      <td>38-40%</td>
      <td>24-28%</td>
    </tr>
    <tr>
      <td>Miso</td>
      <td>Rice - 83-90% <br />Barley - 50-90% (pearled)</td>
      <td>28-30%<br />28-30%</td>
      <td>36-37%<br />38-40%</td>
      <td>20%</td>
    </tr>
    <tr>
      <td>Hatcho miso &amp;<br />Miso-dama</td>
      <td>Soybeans - N/A</td>
      <td>60%</td>
      <td>65%</td>
      <td>20-40%</td>
    </tr>
    <tr>
      <td>Shochu**</td>
      <td>Rice - TBD<br />Barley - TBD</td>
      <td>28-30%<br />28-30%</td>
      <td>38-42%</td>
      <td>17-22%</td>
    </tr>
    <tr>
      <td>Shoyu*<br />(water content)</td>
      <td>Soybeans - N/A<br />Wheat - N/A</td>
      <td>60%* <br />13%* (not soaked)</td>
      <td>58-64%* <br />4% (after roasting)*</td>
      <td>24-28%*</td>
    </tr>
  </tbody>
</table>

<p>*For shoyu, the true water content is shown. It seems brewers prefer to use true water content for shoyu. Some of the shoyu values need to be confirmed. The values for shoyu depend on the soybean to wheat ratio.</p>

<p>**For shochu, some values need to be confirmed. There are additional substrates not listed here. There may also be a difference between the two grades of shochu (continuously distilled 甲類焼酎 and single distilled 乙類焼酎) - to be confirmed.</p>

<h4 id="glossary-of-japanese-terms">Glossary of Japanese Terms</h4>

<p>Here is a list of Japanese terms pertaining to the koji-making process. This can be used in conjunction with the reference temperature curves. Some terms were not covered in this post - but I’ve explained them briefly here.</p>

<table class="table-clean">
  <thead>
    <tr>
      <th>Literal English / Note</th>
      <th>Japanese</th>
      <th>Romaji  / Pronunciation</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>Substrate temperature</td>
      <td>品温</td>
      <td>Hin-on</td>
    </tr>
    <tr>
      <td>Lit.  ‘pull in’ - pull steamed substrate into koji room / machine</td>
      <td>引込み</td>
      <td>Hiki-komi</td>
    </tr>
    <tr>
      <td>Lit.  ‘cut with spores’ - inoculation</td>
      <td>種切り</td>
      <td>Tane-kiri</td>
    </tr>
    <tr>
      <td>Lit.  ‘put on table’ - heaping on table for germination</td>
      <td>床もみ</td>
      <td>Toko-momi</td>
    </tr>
    <tr>
      <td>Lit.  ‘sideburns’ - germination, often in reference to temperature</td>
      <td>もみ上げ</td>
      <td>Momiage</td>
    </tr>
    <tr>
      <td>Lit.  ‘cutting back’ - action of mixing, and also the first mixing step</td>
      <td>切返し</td>
      <td>Kiri-kaeshi</td>
    </tr>
    <tr>
      <td>Lit.  ‘peak’ - transfer to lid / box / etc.</td>
      <td>盛り</td>
      <td>Mori</td>
    </tr>
    <tr>
      <td>Lit.  ‘maintenance’ - non-specific mixing step</td>
      <td>手入れ</td>
      <td>Teire</td>
    </tr>
    <tr>
      <td>Lit.  ‘middle work’ - middle mixing step</td>
      <td>仲仕事</td>
      <td>Naka-shigoto</td>
    </tr>
    <tr>
      <td>Lit.  ‘transhipment’ - shuffling lids or boxes in lid / box method</td>
      <td>積替え</td>
      <td>Tsumi-kae</td>
    </tr>
    <tr>
      <td>Lit.  ‘final work’ - last mixing step</td>
      <td>仕舞仕事</td>
      <td>Shimai-shigoto</td>
    </tr>
    <tr>
      <td>Lit.  ‘final transhipment’</td>
      <td>最高積替え</td>
      <td>Saikō-tsumi-kae</td>
    </tr>
    <tr>
      <td>Lit.  ‘final temperature’ - highest reached temperature in sake-koji</td>
      <td>最高温度</td>
      <td>Saikō-ondo</td>
    </tr>
    <tr>
      <td>Lit.  ‘exit koji’ - koji removed from room</td>
      <td>出麹</td>
      <td>De-kōji</td>
    </tr>
    <tr>
      <td>Young koji - favouring fragrance</td>
      <td>若麹</td>
      <td>Waka-kōji</td>
    </tr>
    <tr>
      <td>Old koji - favouring enzymes</td>
      <td>老ね麹</td>
      <td>Hine-kōji</td>
    </tr>
    <tr>
      <td>Conditioning / drying period after dekoji</td>
      <td>枯らし</td>
      <td>Karashi</td>
    </tr>
    <tr>
      <td>Haze - description of koji growth (hiragana sometimes used)</td>
      <td>破精（はぜ）</td>
      <td>Haze</td>
    </tr>
    <tr>
      <td>Surface haze</td>
      <td>破精廻り</td>
      <td>Haze-mawari</td>
    </tr>
    <tr>
      <td>Interior haze</td>
      <td>破精込み</td>
      <td>Haze-komi</td>
    </tr>
    <tr>
      <td>Lit.  ‘missed’ haze - low surface &amp; interior growth</td>
      <td>破精落ち</td>
      <td>Haze-ochi</td>
    </tr>
    <tr>
      <td>Lit.  ‘painted’ haze - good surface, bad interior growth</td>
      <td>ぬりはぜ</td>
      <td>Nuri-haze</td>
    </tr>
    <tr>
      <td>Lit.  ‘speckled’ haze - low surface, good interior growth</td>
      <td>突はぜ</td>
      <td>Tsuki-haze</td>
    </tr>
    <tr>
      <td>Lit.  ‘total’ haze - good surface, good interior growth</td>
      <td>総はぜ</td>
      <td>Sou-haze</td>
    </tr>
    <tr>
      <td>Lit.  ‘fool’ haze - same as sou-haze but too much moisture</td>
      <td>ばかはぜ</td>
      <td>Baka-haze</td>
    </tr>
    <tr>
      <td>Rice polishing ratio</td>
      <td>精米歩合</td>
      <td>Seimai buai</td>
    </tr>
    <tr>
      <td>Water absorption ratio (weight ratio)</td>
      <td>吸水歩合</td>
      <td>Kyūsui buai</td>
    </tr>
    <tr>
      <td>Soaked rice water absorption ratio (weight ratio)</td>
      <td>浸漬米吸水率</td>
      <td>Shinshimai kyuusui-ritsu</td>
    </tr>
    <tr>
      <td>Steamed rice water absorption ratio (weight ratio)</td>
      <td>蒸米吸水率</td>
      <td>Mushimai kyuusui-ritsu</td>
    </tr>
    <tr>
      <td>Dekoji ratio (weight ratio)</td>
      <td>出麹歩合</td>
      <td>De-kōji buai</td>
    </tr>
  </tbody>
</table>

<h4 id="references">References</h4>

<ol>
  <li>Murakami Hideya. Koji Studies - 6th edition. The Brewing Society of Japan: Tokyo, 2018.</li>
  <li>Katsuya Gomi, Regulatory mechanisms for amylolytic gene expression in the koji mold Aspergillus oryzae. Bioscience, Biotechnology, and Biochemistry 83:8 (2019), 1385-1401, https://doi.org/10.1080/09168451.2019.1625265</li>
  <li>Shigetoshi Sudo et al., Factors in the formation of haze. Journal of The Brewing Society of Japan, 97:5 (2002), 369-376, https://doi.org/10.6013/jbrewsocjapan1988.97.369</li>
  <li>Kazunari Ito et al., Quantitative evaluation of <em>haze</em> formation of <em>koji</em> and progression of internal <em>haze</em> by drying of <em>koji</em> during <em>koji</em> making. Journal of Bioscience and Bioengineering, 124:1 (2017), 62-70, https://doi.org/10.1016/j.jbiosc.2017.02.011</li>
  <li>Wisman, A. P et al., Mapping haze-komi on rice koji grains using b-glucuronidase expressing Aspergillus oryzae and mass spectrometry imaging. Journal of Bioscience and Bioengineering, 129:3 (2020), 269-301, https://doi.org/10.1016/j.jbiosc.2019.09.016</li>
</ol>]]></content><author><name>keith</name></author><category term="koji" /><category term="koji" /><category term="solid-state-fermentation" /><category term="temperature" /><summary type="html"><![CDATA[Koji generates a considerable amount of heat as it grows - if you’ve ever made koji at home, it’s almost unsettling to feel the warmth of a koji cake. Controlling that heat, along with water content, mixing, and ventilation, is crucial to produce a quality product. In this article, we’ll dive into how temperature and other factors are both controlled and exploited in the koji-making process.]]></summary><media:thumbnail xmlns:media="http://search.yahoo.com/mrss/" url="https://controlledmold.com/assets/images/koji_temp_curve/header.png" /><media:content medium="image" url="https://controlledmold.com/assets/images/koji_temp_curve/header.png" xmlns:media="http://search.yahoo.com/mrss/" /></entry><entry><title type="html">Building a bioreactor, part 3: tracking growth rates in real time</title><link href="https://controlledmold.com/building-a-bioreactor-part-3-tracking-growth-rates-in-real-time/" rel="alternate" type="text/html" title="Building a bioreactor, part 3: tracking growth rates in real time" /><published>2020-11-25T00:00:00+00:00</published><updated>2020-11-25T00:00:00+00:00</updated><id>https://controlledmold.com/building-a-bioreactor-part-3-tracking-growth-rates-in-real-time</id><content type="html" xml:base="https://controlledmold.com/building-a-bioreactor-part-3-tracking-growth-rates-in-real-time/"><![CDATA[<p>Part of our bioreactor project depends on the ability to track microbial population sizes. If you read our last blog <a href="https://controlledmold.com/building-a-bioreactor-vessel-and-structure/">article</a>, you’ll know that we measure the population size by the amount of light scattering off the cells and hitting a photodiode. Therefore more light hitting the photodiode ⇒ more cells (up to a limit, but we can ignore that for now). However, this relationship is only proportional, and unless we calibrate our light readings against known cell densities, we don’t know the exact cell density. We’ll see that this isn’t a big deal later. But just know that we are working with optical density only.</p>

<p>If the microbes are happy and healthy, they will be replicating, and our optical density will increase exponentially, that is,</p>

\[\text{OD}_t = \text{OD}_0 \exp{(r\cdot t)}\]

<p>And, after taking measurements, this is what we get:</p>

<p><img src="/assets/images/bioreactor_project/growth_rates/1.png" alt="" /></p>

<p>That looks exponential, right? Well, it is and it isn’t. Let’s first test an exponential by fitting this curve to an exponential model using least-squares:</p>

<div class="language-plaintext highlighter-rouge"><div class="highlight"><pre class="highlight"><code>from scipy.optimize import curve_fit

def exp_(x, A, rate):
    return A*np.exp(rate * x)

results, _ = curve_fit(exp_, x, y, p0=[0.2, .01])
</code></pre></div></div>

<p><img src="/assets/images/bioreactor_project/growth_rates/2.5.png" alt="" /></p>

<p>Wait - what happened? Why is the fit so poor? Let’s try just fitting to the last few hours:</p>

<p><img src="/assets/images/bioreactor_project/growth_rates/2.png" alt="" /></p>

<p>That looks good in the last few hours, but fails everywhere else. Likewise, if we fit to the beginning:</p>

<p><img src="/assets/images/bioreactor_project/growth_rates/3.png" alt="" /></p>

<p>Yuck. What’s going on? The problem is that our growth rate parameter is not constant, so there is no good single value of $r$.  Our model should in fact be:</p>

\[\text{OD}_t = \text{OD}_0 \exp{(r_t\cdot t)}\]

<p>Note the subscript $t$ on the growth rate, $r$. This makes biological sense. Individuals cells, due to variation, will start to reproduce at different times, leading to an increasing population growth rate. Eventually, the growth rate will level off as all cells are actively reproducing (and, given a long enough time, the growth rate will creep up due to evolution). Now that we have a better mathematical model of our population, how can we estimate $r_t$?</p>

<p>To solve this, we will use a Kalman Filter. A Kalman Filter is an algorithm used for <em>estimating</em> a <em>time-varying process</em> that contains an internal <em>dynamical system</em>. Let’s break that down: <em>estimating</em> is the act of taking in noisy measurements and producing a more accurate result; <em>time-varying process</em> is an object that changes over time (in our case: both optical density and growth rate change over time); <em>dynamical system</em> is a model to describe the relationship between our variables: optical density and growth rate. I’d love to go into detail about how a Kalman Filter works, but for a computation-first-mathematics-second resource, I suggest the book I used: <a href="https://github.com/rlabbe/Kalman-and-Bayesian-Filters-in-Python">Kalman and Bayesian Filters in Python</a>.</p>

<p>To start, we’ll write down the variables we wish to estimate (these are called our state variables):</p>

\[\text{OD}_t\]

\[r_t\]

<p>We can also write down a mathematical model of how these are related to each other and <em>their previous values</em>:</p>

\[\begin{align*}
\text{OD}_t &amp;= \text{OD}_{t-1}\exp{(r_{t-1} \cdot \Delta t)}\\
r_{t} &amp;= r_{t-1}
\end{align*}\]

<p>That is, the next optical density is the previous optical density, multiplied by some growth rate. The growth rate is the same as the previous growth rate. Wait - <em>the growth rate is the same as the previous growth rate?</em> I thought we wanted to model a time-varying growth rate? We do, and these equations aren’t complete. There’s an additional noise term:</p>

\[\begin{align*}
\text{OD}_t &amp;= \text{OD}_{t-1}\exp{(r_{t-1} \cdot \Delta t)} + \text{noise}\\
r_{t} &amp;= r_{t-1} + \text{noise}
\end{align*}\]

<p>We usually don’t write the noise terms in the dynamical system, but they are there. The noise terms will allow the rate term to evolve over time.</p>

<p>Without getting further into the math, we can test using a Kalman Filter on our problem above. At the end, we should have a “filtered” version of our optical density (think of it as smoothed - the noise has been eliminated), and a time-varying estimate of our growth rate.</p>

<p><img src="/assets/images/bioreactor_project/growth_rates/4.png" alt="" /></p>

<p><img src="/assets/images/bioreactor_project/growth_rates/5.png" alt="" /></p>

<p>Looking at the second figure, the estimated growth rate, we can see that indeed the growth rate is increasing over time. It is always positive, which means the population will still be exponentially increasing at all times (as observed). However, we see some “noise” in it - what are those periodic bumps in the right of the graph? If we go back to the first figure in the article, we can see little blips in the optical density - our growth rate estimate is picking these up. It turns out, the room’s heating turning on is causing these bumps. It’s cool that we can pick these minor disturbances up.</p>

<p>This is great - we now have the following:</p>

<ol>
  <li>A time-varying estimate of growth rate (as we expect biological populations to have a time-varying growth rate).</li>
  <li>The nature of the estimation also allows us to perform this in real time. That is, given a new OD measurement, we can pass it into our Kalman Filter and produce a new estimate of the growth rate at that time. This means we have immediate data for decision making. Compare this to the traditional “batch” process: take the past 30 minutes of data, run it through a curve-fitter, produce an estimate. You might say: well this works, too - if our 30 minute window is sliding, we can always produce an up-to-the-second growth rate estimate from the past 30 minute window. Here are some problems:
    <ul>
      <li>Why 30 minutes? Why not 10 minutes, or 1 minute? There’s a bias-variance tradeoff to make here.</li>
      <li>If the growth rate is changing quickly, your window may be too long and you’ll miss the movements.</li>
      <li>What happens when there is a dilution event?</li>
    </ul>
  </li>
</ol>

<h3 id="dilution-events">Dilution events</h3>
<p>What is a dilution event? Well, the nature of exponential growth is that it will grow without bound. Obviously there is an upper limit based on the amount of nutrients, but we would like to keep our culture in a state of exponential growth. This requires is to periodically expel some liquid, and add in fresh media. This causes a reduction of the overall cell density, and therefore a reduction in the optical density too. You can see this in the figure below. The dips on the right-hand side is us removing some fraction of liquid and replacing it with fresh media. In fact, this is on a scheduale, and we perform this dilution every 45 minutes.</p>

<p><img src="/assets/images/bioreactor_project/growth_rates/6.png" alt="" /></p>

<p>From the point-of-view of our Kalman Filter, it will get confused about these drops in OD: it would say “wowow lots of negative growth!” - but we know that’s not true. The growth rate shouldn’t vary over a dilution event - the change in OD is due to a physical process, not a biological one.</p>

<p>Luckily, Kalman Filters can be “hacked” to include external information. I’ve programmed just this (taking inspiration from <a href="https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0181923#pone.0181923.s007">this paper</a>). Basically, during the dilution events, we increase the noise many fold, causing all the estimates to “freeze”, until the noise is back down when the dilution event ends.</p>

<p><img src="/assets/images/bioreactor_project/growth_rates/7.png" alt="" /></p>

<p>What’s up with the oscillations on the right? Are those artifacts of our hack? Actually no, it’s a real phenomenon. Zooming in on the dilution events:</p>

<p><img src="/assets/images/bioreactor_project/growth_rates/8.png" alt="" /></p>

<p>We can see that the growth rate is not constant between dilution events, and does vary up then down. Our Kalman Filter (for better or worse) picks this up.</p>

<h3 id="sensor-fusion">Sensor fusion</h3>

<p>Using a Kalman Filter allows us to combine multiple sensors into a single estimate. For example, in our bioreactor, we actually have multiple photodiodes placed at different angles. Different angles have different sensitivities. We would like some way to combine all these photodiode measurements together. Ideally, this new estimate should have a smaller variance than estimates provided from not combining measurements (spoiler alert: it does have smaller variance). The different sensors don’t even have to be the same type. Suppose I have a carbon dioxide sensor in the headspace. As the cells grow, they will produce more and more CO₂. If I have a model of how CO₂ is related to the growth rate, I can include this (possibly very noisy) measurement and get a more accurate growth rate estimate.</p>

<h3 id="modeling-saturation">Modeling saturation</h3>

<p>I mentioned above that the light scattering is proportional to cell density <em>up to a limit</em>. That limit is called the saturation point. Beyond this point, the relationship is no longer linear, but sub-linear. This is because there are so many cells that light that would previously have scattered into the photodiode is now being scattered away. There are atleast two solutions:</p>

<ol>
  <li>Use angles that have a higher saturation point, like 90°. This comes at a cost of sensitivity though.</li>
  <li>Model the saturation in our dynamical system.</li>
</ol>

<p>The latter might look like:</p>

\[\begin{align*}
C_t &amp;= C_{t-1} \exp{(r_{t-1} \cdot \Delta t)}\\
\text{OD}_t &amp;= f(\text{C}_{t-1}, \theta)\\
r_{t} &amp;= r_{t-1}
\end{align*}\]

<p>where $f$ is some linear function up until $\theta$, our saturation limit, and sub-linear beyond that, and $C_t$ is a (hidden) state of cell population. The $\theta$ can be estimated in the Kalman Filter, too. I haven’t implemented this yet, though!</p>

<h3 id="conclusion">Conclusion</h3>

<p>To be honest, I have a love-hate relationship with Kalman filers. They enable some amazing  inferences, like in our example above, that would otherwise be unreachable. However, they come with lots of parameters that need tuning. This has been a pain for me, as it’s not always clear how the model will change with respect to changing a parameter. The book I mentioned above, <a href="https://github.com/rlabbe/Kalman-and-Bayesian-Filters-in-Python">Kalman and Bayesian Filters in Python</a>, has certainly helped with this, though.</p>

<p>Now that we have a good estimate time-varying growth rate, we monitor how the culture reacts to different environments. By controlling the environment, we can control the growth rate. So by <em>surpressing</em> the growth rate using the environment, we are introducing “headspace” for the microbes to evolve into. We’ll explore this next article.</p>]]></content><author><name>cam</name></author><category term="data-science" /><category term="bioreactor-project" /><category term="bioreactor" /><category term="IoT" /><category term="sensors" /><category term="data-science" /><summary type="html"><![CDATA[Part of our bioreactor project depends on the ability to track microbial population sizes. If you read our last blog article, you’ll know that we measure the population size by the amount of light scattering off the cells and hitting a photodiode. Therefore more light hitting the photodiode ⇒ more cells (up to a limit, but we can ignore that for now). However, this relationship is only proportional, and unless we calibrate our light readings against known cell densities, we don’t know the exact cell density. We’ll see that this isn’t a big deal later. But just know that we are working with optical density only.]]></summary><media:thumbnail xmlns:media="http://search.yahoo.com/mrss/" url="https://controlledmold.com/assets/images/bioreactor_project/growth_rates/header.png" /><media:content medium="image" url="https://controlledmold.com/assets/images/bioreactor_project/growth_rates/header.png" xmlns:media="http://search.yahoo.com/mrss/" /></entry><entry><title type="html">Building a bioreactor, part 2: Vessel and structure</title><link href="https://controlledmold.com/building-a-bioreactor-vessel-and-structure/" rel="alternate" type="text/html" title="Building a bioreactor, part 2: Vessel and structure" /><published>2020-10-20T00:00:00+00:00</published><updated>2020-10-20T00:00:00+00:00</updated><id>https://controlledmold.com/building-a-bioreactor-vessel-and-structure</id><content type="html" xml:base="https://controlledmold.com/building-a-bioreactor-vessel-and-structure/"><![CDATA[<p>The first part of building our bioreactor will focus on the container for the glass vial. The original designs for this are from another open-source project (and research article), <a href="https://www.biorxiv.org/content/10.1101/729434v1">“A low-cost, open source, self-contained bacterial EVolutionary biorEactor (EVE)”</a>. 3D sketch below!</p>

<p><img src="/assets/images/bioreactor_project/vessel/v2_tri.jpg" alt="/assets/images/bioreactor_project/vessel/v2_tri.jpg" class="responsive center" /></p>

<p>We’ve made some changes to it, which I’ll get to later. For now, I want to explain the mechanism of how it works. A 20ml vial fits snuggle in the center:</p>

<p><img src="/assets/images/bioreactor_project/vessel/vessel2.jpg" alt="/assets/images/bioreactor_project/vessel/vessel2.jpg" class="responsive center" /></p>

<p>The holes, angled at 135° from each other, are designed to fit LEDs and a cap goes over the exposed end.</p>

<p><img src="/assets/images/bioreactor_project/vessel/led.jpeg" alt="/assets/images/bioreactor_project/vessel/led.jpeg" class="responsive center" /></p>

<p>What LEDs though? In one hole, we fit an infrared (IR) <em>LED</em>, and in the other hole we fit an infrared <em>photodiode</em>. The latter is a component that produces a current when IR light hits it. This is how we will measure the turbidity (the cloudiness) of the solution, which is a measure of cell density!</p>

<p>The theory is that cells will scatter the IR light off their surfaces and more cells ⇒ more scatter ⇒ more light picked up by the photodiode. In fact, the relationship is linear (up to a limit), so current produced should be proportional to cell density. Later, in another article, we will learn how to translate that current into a computer readable signal.</p>

<p><img src="/assets/images/bioreactor_project/vessel/scatter.png" alt="/assets/images/bioreactor_project/vessel/scatter.png" class="responsive center" /></p>

<p>It’s great that we can measure cell density, but what if we also wanted to build a photobioreactor? That is, what if we wanted to add LEDs that phototrophic algae could feed off of? Similarly, what if we wanted to introduce UV light into the culture, to induce higher rates of mutation in the organims? Furthermore, if we have <em>more</em> IR photodiodes, we could get better statistical estimates in our inference pipeline. Introducing our new vessel:</p>

<p><img src="/assets/images/bioreactor_project/vessel/vesselv2.jpg" alt="/assets/images/bioreactor_project/vessel/vesselv2.jpg" class="responsive center" />
<em>Yes, this has been printed and is running now. 3D printer schematics available <a href="https://github.com/CamDavidsonPilon/morbidostat/blob/master/3D_files/Holder%20v4.stl">here</a></em></p>

<p>This has six LED holders for all the applications listed above. In the image below, we show off the application of 4 photodiodes (at 180°, 135°, 90° and 45°) and a UV LED.</p>

<p><img src="/assets/images/bioreactor_project/vessel/scatter_diagram_(1).png" alt="/assets/images/bioreactor_project/vessel/scatter_diagram_(1).png" class="responsive center" /></p>

<p>The different angles have different uses: the 180° can measure absorbance, and the others can measure scatter. The 135° has the strongest signal (due to the most scattering in that direction), however has the lowest saturation (too many cells ⇒ too much turbidity ⇒ loss of signal). The 90° and 45° photodiodes have a weaker signal, but will work better in high-turbidity environments. This really matters: take a look at the following data:</p>

<p><img src="/assets/images/bioreactor_project/vessel/Untitled.png" alt="/assets/images/bioreactor_project/vessel/Untitled.png" class="responsive center" />
*The growth curves respresented by a 135° sensor and a 90° sensor. The 135° sensor saturates quicker due to high turbidity. The noise at 16hours in was caused by me playing with wires. *</p>

<p>Both start off with the same pattern: exponential growth, but the 135° photodiode tapers off more quickly, while the 90° photodiode continues to grow and exhibit a more “classic” growth curve.</p>

<h3 id="magnetic-stirring">Magnetic stirring</h3>

<p>Below each holder is a stand with a small 40mm DC fan. Attached to to the fan blades are two small magnets. When the fan is turned on, the magnets spin a corresponding stir bar in the vial above. he fan is powered by a MOSFET, and the speed is controlled by a pulse-width-modulation signal from a RaspberryPi. More on that in a future article!</p>

<p><img src="/assets/images/bioreactor_project/vessel/spin.gif" alt="/assets/images/bioreactor_project/vessel/spin.gif" class="responsive center" />
<em>slow-motion video of our magnets and fan</em></p>]]></content><author><name>cam</name></author><category term="bioreactor-project" /><category term="bioreactor" /><category term="IoT" /><category term="sensors" /><category term="submerged-fermentation" /><summary type="html"><![CDATA[The first part of building our bioreactor will focus on the container for the glass vial. The original designs for this are from another open-source project (and research article), “A low-cost, open source, self-contained bacterial EVolutionary biorEactor (EVE)”. 3D sketch below!]]></summary><media:thumbnail xmlns:media="http://search.yahoo.com/mrss/" url="https://controlledmold.com/assets/images/bioreactor_project/vessel/1.png" /><media:content medium="image" url="https://controlledmold.com/assets/images/bioreactor_project/vessel/1.png" xmlns:media="http://search.yahoo.com/mrss/" /></entry><entry><title type="html">Temperature, water content, and polishing effects on koji</title><link href="https://controlledmold.com/temperature-water-content-and-polishing-effects-on-koji/" rel="alternate" type="text/html" title="Temperature, water content, and polishing effects on koji" /><published>2020-10-10T00:00:00+00:00</published><updated>2020-10-10T00:00:00+00:00</updated><id>https://controlledmold.com/temperature-water-content-and-polishing-effects-on-koji</id><content type="html" xml:base="https://controlledmold.com/temperature-water-content-and-polishing-effects-on-koji/"><![CDATA[<p>I always wondered if researchers had played around with koji growing parameters - turns out they have, and with a great level of detail in numerous studies. This post will be the first of a few going over these studies - starting with temperature, water content, and polishing rate.</p>

<h3 id="overview">Overview</h3>

<p>Koji-making is all about optimizing growth conditions for the specific application you’re after. Although koji-makers had centuries of knowledge about optimal growth conditions, it wasn’t examined  scientifically until the 20th century.</p>

<p>Everything in this post comes from the Brewing Society of Japan, and its academic journal (which has been running from 1906 to today)  - more on that at the end of the post. The experiments below were run mostly in the the 1970s and the 1980s by researchers looking to explain and quantify different aspects of koji-making.</p>

<h3 id="temperature">Temperature</h3>

<p>One of the most obvious parameters that researchers tested was the <strong>cultivation temperature</strong>. The growth of A. oryzae on white rice was tested at temperatures between 30°C and 40°C. Of course, the researchers know that growing koji for a fixed time at different temperatures would not be a fair comparison: at different temperatures, the koji’s growth rate (μ, measured in h<sup>-1</sup>) is different, and hence enzyme production is different. Thus, enzyme production was compared at two points different points:</p>

<ol>
  <li>When the koji consumes 10mL oxygen per gram of koji (dry weight basis), and;</li>
  <li>When the koji reaches a maximum cell mass</li>
</ol>

<p><img src="/assets/images/temp_effect/Table 1.png" alt="Table 1" />
<em>Effect of culture temperature on enzyme production ([1] - 10mL O<sub>2</sub> consumed/ g koji and [2] at maximum cell mass). Units are arbitrary, but higher values imply more activity in the assay.</em></p>

<p><em>Conclusion:</em> in general, higher temperatures favour amylases, and lower temperatures favour proteases and peptidases.</p>

<p>In practice, no koji makers would set a single temperature for their koji for the entire duration of its growth. The above temperatures are mostly important during the last 12 hours of growth when the mycelium is producing the most enzymes.</p>

<p>As an aside, this was the experimental set-up:</p>

<p><img src="/assets/images/temp_effect/Figure 1.png" alt="Figure 1" class="responsive center" />
<em>The experimental setup - see Ref. 2</em></p>

<div class="callout">
💡 This setup is a sort of respirometer - I'm not sure of the exact name. Oxygen is produced by electrolysis, which builds up pressure in the system, pushing one side of the mercury manometer down. With enough pressure, the mercury level switch opens, stopping the generation of oxygen. The koji continues consuming oxygen, and one can measure total oxygen consumption by integrating the total current consumed over time. Finally, waste carbon dioxide is absorbed by the NaOH solution.
</div>

<h3 id="water-content">Water content</h3>

<p>Another parameter they tested is the <strong>steamed rice water absorption rate</strong>, which is defined as follows:</p>

\[\text{Steamed rice water absorption rate (%)} = \frac{\text{Steamed rice weight} - \text{Dry rice weight}}{\text{Dry rice weight} }   \times 100\%\]

<p>Note that this is NOT the same as total water content, since the dry white rice also contains a small amount of water (around 15% by weight). The term “dry rice weight” here is a bit of a misnomer - it’s the weight of uncooked, unsoaked rice. When varying the steamed rice water absorption rate from 24 - 51%, this is what they found:</p>

<p><img src="/assets/images/temp_effect/Table 2.png" alt="Table 2" />
<em>Effect of water absorption rate on enzyme activity</em></p>

<p>The researchers used the same metric above, and pulled the koji once it had consumed 10mL oxygen / g koji for a fair comparison. Here, the total culture time is an indicator of growth rate: the shorter the culture time, the faster the growth rate.</p>

<p><em>Conclusion:</em> although wetter rice will grow koji faster, it yields less enzymes in total (both amylase and proteases).</p>

<p>Koji makers already knew that overly wet rice yielded weak koji. Another researcher found that wet rice lead to the over-accumulation of glucose, which inhibits enzyme production. In practice, the water content actually changes throughout the koji production period as some of the water vapour is released from the rice. Typical operations start at 35-40% steamed rice water absorption rate, and finish at 24 - 28% (mirin) down to 17-19% (sake).</p>

<h3 id="polishing-rate">Polishing rate</h3>

<p>Yet another parameter that was tested is the <strong>polishing rate</strong>, which is defined as follows:</p>

\[\text{Polishing rate (%)} = \frac{ \text{Weight of white (polished) rice}}{\text{Weight of brown rice}} \times 100\%\]

<p>Although a bit counter-intuitive, a <em>lower</em> polishing rate means <em>more</em> of the rice is removed.  Polishing rates were varied from 65% - 95% on three varieties of rice.</p>

<p><img src="/assets/images/temp_effect/Table 3.png" alt="Table 3" />
<em>Effect of polishing rate on enzyme activity</em></p>

<p>Again, the koji was pulled once it had consumed 10mL oxygen / g koji.</p>

<p><em>Conclusion:</em> rice variety doesn’t matter much - and enzyme yields are mixed. Rice with a high polishing rate, which contains more nutrients, grows koji faster. Rice with a low polishing rate, which is mostly starch, grows slower.</p>

<p>In practice, miso and mirin typically use rice with a high polishing rate. Rice with lower polishing rates are reserved for higher and higher grades of sake - where depriving the sake mash of nutrients yields sake with a clean, refined taste.</p>

<h3 id="other-effects">Other effects</h3>

<p>There are of course, many more parameters that were tested. Here are some of them:</p>

<ol>
  <li>Carbon dioxide</li>
  <li>Oxygen</li>
  <li>Air humidity</li>
  <li>Inoculation rate</li>
  <li><strong>The addition of salts &amp; minerals</strong></li>
</ol>

<p>This last one, the addition of salts &amp; minerals, is really interesting and needs another post. As for the other four, in short - they don’t significantly affect koji quality within practical bounds. If there’s enough interest, I’ll look at them in detail.</p>

<h3 id="conclusions">Conclusions</h3>

<h4 id="notes">Notes</h4>

<p>I’ve omitted units for enzyme activity, which are described in the paper, along with the experimental setup and methodology. I’ve also omitted results for Tyrosinase and Deferriferrichrysin, which were also measured in these studies.</p>

<h4 id="the-brewing-society-of-japan">The Brewing Society of Japan</h4>

<p>A quick footnote: a lot of the information in this post comes from the <a href="https://www.jozo.or.jp/en/about/">Brewing Society of Japan</a>, which was established in 1906. Their articles are availble on <a href="https://www.jstage.jst.go.jp/browse/jbrewsocjapan/-char/en">J-Stage</a>. There are also a other useful journals like the <a href="https://www.jstage.jst.go.jp/browse/nskkk/-char/ja">Journal of Japanese Society of Food Science and Engineering</a>.</p>

<p>For the Brewing Society of Japan, they also have a <a href="https://www.jozo.or.jp/booksales/">Bookstore</a>, and the books help distill all the information from the Journal. Notable titles are Koji Studies「麹学」and Molecular Koji-kin Studies 「分子麹菌学」</p>

<p>I’m trying to translate as much these as I can, but if you’re interested, message me on twitter @Keeifa</p>

<h4 id="additional-reading">Additional reading</h4>

<p>Rice polishing ratio, explained by <a href="http://www.nada-ken.com/main/en/index_s/84.html">Nada-ken</a></p>

<h4 id="references">References</h4>

<p>Murakami Hideya. Koji Studies - 6th edition. Chapter 5 - Koji Enzymes. The Brewing Society of Japan: Tokyo, 2018.</p>

<p>Naoto Okazaki et al. Effects of koji-making conditions on growth and enzyme production. Journal of The Brewing Society of Japan, 74:10 (1979), 683-686, https://doi.org/10.6013/jbrewsocjapan1915.74.683</p>]]></content><author><name>keith</name></author><category term="koji" /><category term="koji" /><category term="amylase" /><category term="enzymes" /><category term="rice" /><category term="water-content" /><category term="polishing" /><category term="temperature" /><summary type="html"><![CDATA[I always wondered if researchers had played around with koji growing parameters - turns out they have, and with a great level of detail in numerous studies. This post will be the first of a few going over these studies - starting with temperature, water content, and polishing rate.]]></summary><media:thumbnail xmlns:media="http://search.yahoo.com/mrss/" url="https://controlledmold.com/assets/images/temp_effect/header.png" /><media:content medium="image" url="https://controlledmold.com/assets/images/temp_effect/header.png" xmlns:media="http://search.yahoo.com/mrss/" /></entry></feed>