Fully Automated Brewery Senior Design

jpalarchio said:

The one place a solenoid valve might make sense over a ball valve is if you're going to have one for water fill.

The solenoid valve shuts off immediately versus the 10-15 seconds a ball valve takes to close.

Click to expand...

That is somehting I have though about in the past... my thoughts was to close the valve as you approach the set point - if you close it half way you should be able to slow fill the last bit and not overshoot much when you finally shut the valve off. Kind of like a petrol bowser.

mattd2 said:

That is somehting I have though about in the past... my thoughts was to close the valve as you approach the set point - if you close it half way you should be able to slow fill the last bit and not overshoot much when you finally shut the valve off. Kind of like a petrol bowser.

Click to expand...

The ball valves I've seen and used were not proportional, at least not anything I've seen that's in my budget. I think a few people have tried to modify various valves but they're basically open or closed. Usually this comes up when people are trying to control sparge rates with an automated ball valve.

jpalarchio said:

The ball valves I've seen and used were not proportional, at least not anything I've seen that's in my budget. I think a few people have tried to modify various valves but they're basically open or closed. Usually this comes up when people are trying to control sparge rates with an automated ball valve.

Click to expand...

Sorry, yes the typical ones sold by OSCsys are fail open (or close????) and typically wired with a constant supply and a control input for close (or open????). What I was suggesting is to also control the constant supply. If they are actually sprung closed or ahve a capacitor in them to provide power in case of failure then there would be more to it. For what it is worth you can purchase from Asia with more control options - one being controlled open/close with fully open/closed limit siwthces!
http://www.tf-valvefittings.com/goods-13.html - wiring diagram CR05
When I said close 50% I mean roughly close it for 5 seconds.

I've solved the valve PID workaround by using 2 things:
1) the motors will be controlled with either PWM or increasing/ decreasing DC voltage depending on the motor.
2) flow rate can be calculated two ways that I've thought of so far
2a) Integration of an analog flow sensor
2b) knowing the total volume of both the hlt and kettle, you can "differentiate" two points in time to find the slope of the volume fill thus giving you flow rate.

Having played with various motors for a year in the lab, its very very difficult to integrate or differentiate [bold]position[/bold] of the rotor without a tachometer. Thus, i believe its much easier and fool proof to buy cheaper electric ball valves (a mechanical device prone to failure) and control flow with motor/pump speed than to rely on a ball valve motor control. Does that make sense?

jgalati said:

Having played with various motors for a year in the lab, its very very difficult to integrate or differentiate [bold]position[/bold] of the rotor without a tachometer. Thus, i believe its much easier and fool proof to buy cheaper electric ball valves (a mechanical device prone to failure) and control flow with motor/pump speed than to rely on a ball valve motor control. Does that make sense?

Click to expand...

I think I am confusing myself by reading both you posts at once and forgetting to differenetiate bewteen the motor(pump) and the motor(valve)

So is my understanding of what you proposed to do correct:
Valves will be on/off only - no flow control
Pump will be flow controlled (in some way yet to be determined depending on the type of pump selected)

From my confusion an idea was born though - if you want to control the flow via a electronic ball valve could you: Select a "sprung closed" type valve and PWM the control signal. Thinking about it PWM at 50% will keep the valve in position, 0%/100% will be "quick" close/open and values between them will slow the opening/closing. Would this not suit a PID control loop well.

It would yes, but controlling a motor is much easier than a ball valve because I'm controlling a constant voltage or PWM rather than polarizing/depolarizing the motor on the ball valve. Also, valves are more prone to failure than motors, and every time a valve is opened and closed it's life shortens. It's also a mechanical device with more moving parts and seals than a pump, which really has 2 moving parts.

jgalati said:

It would yes, but controlling a motor is much easier than a ball valve because I'm controlling a constant voltage or PWM rather than polarizing/depolarizing the motor on the ball valve. Also, valves are more prone to failure than motors, and every time a valve is opened and closed it's life shortens. It's also a mechanical device with more moving parts and seals than a pump, which really has 2 moving parts.

Click to expand...

Man you are really making me want to ditch work and go home and pull out my Arduino, pulse output turbine flow meter, rip my control panel apart to grab the SSR and get a PWM flow control pump with feedback working!

By the way keep focused on then end goal of the project - much easier to write a report on it when you start with a goal and end up at it, rather than start with a goal, end up somehwere else and have to try and come up with a goal that fits within the assignments scope and has the outcomes you ended up with
Keep us updated and this is a great place to bounce ideas around.

Having never brewed on a Herms system, can someone shed insight into the problems they face cleaning their brewery at the end of the day. Also, what's easy to do?

Another question:
Assume whole leaf hops:

How do you guys sanitize your pump and hose before draining kettle to fermenter in herms? Assuming one is using a whirlpool, how do you prevent your mesh screen from becoming clogged before the whirlpool is set? Also, I'd like this project to not use a hop spider. Replacing the paint bag is an unnecessary cost, and from experience, cleaning the bag for reuse is a PITA. A stainless mesh screen would be cool but, again, expensive. Thoughts, opinions, ideas, empirical testing anyone's done? Ideally, one could just throw hops into the kettle.

Update:

I'm working on a simpler design with these assumptions:

1) We'll get CIP working
2) There is already a tap-water solenoid valve for both the boil kettle and HLT (they're only $15)
3) Whirlpool is feasible
4) A stainless steel colander will be in the kettle to block the big hops from getting into the kettle.

Given these, I plan on installing an immersion chiller directly into the boil kettle, and whirl-pooling while cooling. The chiller will be close to the edge of the boil kettle where the velocity of the unfermented beer is greatest, thus increasing the rate of heat transfer.

Here's my rationale:
1) It's cheaper to buy 50' of copper tubing and punch holes in the kettle + connections than it is to build or buy a counterflow chiller, plate chiller, etc.
2) I don't have to worry about sanitizing or cleaning these devices.
3) I'm eliminating more plumbing
4) I'm reducing risk of infection by not having a chiller (terrible reason but I've seen a couple stories).
5) Plumbing is simplified

Anyone ever had a bad experience with an immersion chiller?

At the present, to cool my beer, I use a 25' 3/8" copper chiller with a hose connection outside. I then take a drill with a sanitized paint stirrer, and let that puppy go full blast while water is running. This cools 6.5 gallons of beer in 8' and also aerates it. While I won't get that type of velocity with a whirlpool, I can leave my kettle covered while this process is occurring, giving that most of the DMS have evaporated, and really don't have too much risk of infection.

Thoughts? Thanks.

There are several types of hop blockers/filters that can be used in the kettle. You could simplify the problem by only using leaf hops in the boil and/or using the hop bag for pellet hops. The whirlpool and a course filter should be very effective on the leaf hops.

jgalati said:

Having played with various motors for a year in the lab, its very very difficult to integrate or differentiate [bold]position[/bold] of the rotor without a tachometer. Thus, i believe its much easier and fool proof to buy cheaper electric ball valves (a mechanical device prone to failure) and control flow with motor/pump speed than to rely on a ball valve motor control. Does that make sense?

Click to expand...

Couldn't you turn a ball valve into a servo pretty easily? Even use some of the electronics from a small plastic model servo but replace the motor with a gear motor with a lot of torque. All you need to do is attach a pot to the shaft of the valve.

Spintab said:

Couldn't you turn a ball valve into a servo pretty easily? Even use some of the electronics from a small plastic model servo but replace the motor with a gear motor with a lot of torque. All you need to do is attach a pot to the shaft of the valve.

Click to expand...

I could..... but controlling flow from the pump is easier from a electrical standpoint and is less prone to failure

Subscribed! Awesome and congratulations! Only thing I would add that hasn't been mentioned is touchscreen interface, I know you are trying to keep it on the cheap, but these things have really dropped in price. I think it would be money well spent on the system. Maybe a SQL database to store recipes .. Good luck!

hepkat701 said:

Subscribed! Awesome and congratulations! Only thing I would add that hasn't been mentioned is touchscreen interface, I know you are trying to keep it on the cheap, but these things have really dropped in price. I think it would be money well spent on the system. Maybe a SQL database to store recipes .. Good luck!

Click to expand...

Hey bud. Thanks for subscribing. The interface is going to be on a webappp so all platforms can use the system. The SQL database is actually going to be a drop box folder on your local PC, which has to be connected to the same router as your phone and brewery. The beer XML files will then be grabbed off that folder and displayed on a menu to brew.

If a beer XML doesn't exist for a recipe, the traditional herms methods will be available, I.e. heat strike water, recirculate, boil wort, chill, etc.

Well I can't wait to see this unfold!

hey jgalati,

great idea for a project! wish I would have thought of it when I was in school.

Have you had a look at the Braumeister? Its one of the most ingenuitive devices I've seen so far...very elegant. I could see it being improved with your beerxml idea, along with smartphone app to control it and or load beerxml files for brewing, along with system status updates to the brewer. also, it could easily be fitted with an additional port for a chiller, although that's really not necessary as an immersion chiller is not generally a problem to use. Assuming a fixed install, ports for water lines could be used, and once water lines are installed, a fixed immersion chiller could work (maybe a larger external tank diameter?). The drawbacks of the system is that one needs a second container to heat sparge water, and the sparging is manual (as opposed to an automated fly sparging set up).

I look forward to your progress and success! Good luck!

j

Could you also consider looking into utilising the "in-built" microcontrollers on the Beagle Bones Black for the process control with the main microprocessor handling the UI and communcation tasks.

mattd2 said:

Could you also consider looking into utilising the "in-built" microcontrollers on the Beagle Bones Black for the process control with the main microprocessor handling the UI and communcation tasks.

Click to expand...

I'm currently looking into doing just that. Are you referring to the pin headers?

Sent from my DROID RAZR HD using Home Brew mobile app

jgalati said:

I'm currently looking into doing just that. Are you referring to the pin headers?

Sent from my DROID RAZR HD using Home Brew mobile app

Click to expand...

Refering to the "2x PRU 32-bit microcontrollers" included in the ARM processor. Listed as the 5th bullet point under processor: http://beagleboard.org/products/beaglebone black
Edit: Did a bit more reading - it looks like currently the PRU is programmed from the main CPU in assembler and no cipiler has been created to ease the pain of programming in assmbler (source: http://www.element14.com/community/...3/05/22/bbb--working-with-the-pru-icssprussv2).
Don't quote me as being correct but it also seems like the PRU will also just excute code and you'll have to sort out how to intergrate the addition I/O modules (ADC etc. to them).
Long story short - as part of your project I wouldn't look at it

There's a really nice IDE environment to program in. it's a combination of java and c. definitely thinking of ditching the arduino and using this as a substitute for the raspberry pi and arduino combination

jgalati said:

There's a really nice IDE environment to program in. it's a combination of java and c. definitely thinking of ditching the arduino and using this as a substitute for the raspberry pi and arduino combination

Click to expand...

Can you share a link?
The Arduino has the Tre coming out which if I am correct is a collaboration between Ardunio and TI.

mattd2 said:

Can you share a link?
The Arduino has the Tre coming out which if I am correct is a collaboration between Ardunio and TI.

Click to expand...

http://beagleboard.org/project/bonescript

Sent from my DROID RAZR HD using Home Brew mobile app

I'd rethink the DC pumps, if you are pumping through a HERMS coil. I have center inlet chuggers and 30' of 3/8 coil (I know, I need a bigger coil) and I get less than 2 GPM at full bore. According to the pump curves, that amounts to 17-18 ft of equivalent head loss. I also have as little height difference between pumps and kettles as possible. So the strong majority of the head-loss is from the coil, not height. Those little DC pumps won't get more than a trickle... Low flow through HERSM = low rate of heat exchange. That will make temperature control of the MLT tricky.

You can closed-loop speed control both DC and AC pumps without a position or velocity sensor. I don't mean constant V/Hz operation either, which I wouldn't think would work well in our application. It isn't particularly simple, especially in the DC case. Google for technical papers on sensorless control. It is a HUGE topic in research right now, as sensors are expensive and failure-prone.

At the end of the day, though, it is just easier to get a sensor. A simple encoder, as long as it has 512 or more lines, will be sufficient for speed control.

BadWolfBrewing said:

I'd rethink the DC pumps, if you are pumping through a HERMS coil. I have center inlet chuggers and 30' of 3/8 coil (I know, I need a bigger coil) and I get less than 2 GPM at full bore. According to the pump curves, that amounts to 17-18 ft of equivalent head loss. I also have as little height difference between pumps and kettles as possible. So the strong majority of the head-loss is from the coil, not height. Those little DC pumps won't get more than a trickle... Low flow through HERSM = low rate of heat exchange. That will make temperature control of the MLT tricky.

You can closed-loop speed control both DC and AC pumps without a position or velocity sensor. I don't mean constant V/Hz operation either, which I wouldn't think would work well in our application. It isn't particularly simple, especially in the DC case. Google for technical papers on sensorless control. It is a HUGE topic in research right now, as sensors are expensive and failure-prone.

At the end of the day, though, it is just easier to get a sensor. A simple encoder, as long as it has 512 or more lines, will be sufficient for speed control.

Click to expand...

What GPM do you need for a herms setup?

Sent from my DROID RAZR HD using Home Brew mobile app

thats a tricky question. Depends on a few things I guess. If you have a cooler as a MLT, you can get away with less. You need enough heat exchange to add energy equal to the amount that is being lost to ambient. Any extra heat exchange capability can be used to increase the temperature. As we might step mash, and also raise to mash-out at 168-171, we need enough heat exchange 'head-room' to be able to pump up the temperature at a reasonable rate. I think 2-4 degs F per minute is a great number. I am usually around 2, and thats okay.

The rate of heat exchange is more or less proportional to the flow rate. Some may argue that a faster flow will result in less heat exchange as the wort isn't in contact with the coil as long, but that is nonsensical. That effect will make the relationship a little less than proportional to flow rate, but still monotonic. So, faster flow = more heat exchange.

The limit is either the pumping speed (which is my limit), or grain bed compaction. I've never had that happen, but I am pretty limited in pumping speed. If I was to switch to 50' of 1/2" coil, I should expect at least a 2x increase in maximum flow. That is a tricky thing to calculate, as a flow through a coil is VERY different than a flow through a straight tube. See Dean's number.

At my rate of a little over 1 GPM, grain bed compaction isn't an issue. Also, it takes FOREVER to raise my MLT temp, and if I remove the lid all bets are off. So, faster than 1 GPM. I have stainless stout-tanks vessels, BTW. If it was a cooler, I wouldn't have as much of an issue. They leak energy to ambient MUCH slower, so I don't need to replace energy as quickly.

I've heard of 2 GPM without problems before, so that could be a good number to target. If grain bed compaction occurs, make sure channeling isn't an issue, relax on the crush a bit, or add some rice hulls.

Also, I am of the opinion that convoluted copper, at least in the HERMS context, isn't worth it. Our flow (even my slow flow) is fairly turbulent as it is. Aside from increasing surface area, there shouldn't be much of an impact. And once you are getting close to steady-state temps, even at 2 GPM the HERMS effluence should be so close to the HLT temperature that increasing the heat transfer coefficient wouldn't a difference.

Most folks with a 1/2" 50' coil seem to report no problems in getting the MLT to track the HLT very closely, even through transients. So the convoluted coil doesn't add much. They do look nice though.

Badwolf:
Is it not possible to achieve the same effect, not through flow rate, but by increasing the temperature differential? Should one want to step mash, simply heat the hlt more, and increase the delta T. From a thermo standpoint, what difference is there between 10 feet of tubing with a hotter HLT than 50 feet with a slightly cooler HLT if the exit temp of the mash is the same? By reducing piping, I reduce cost and head from the piping, thereby allowing me to use a smaller pump.

Sent from my DROID RAZR HD using Home Brew mobile app

you are correct. I should have been specific. I close the temperature loop on the HLT temp, under the assumption that the MLT will follow it. I may add a deg or 2, to make up for my little coil.

With a 10' coil, and a big temp difference between the MLT and HLT, I'm not sure that you will get the effluence temp equal to the HLT temp. Doesn't matter though...

If you really want to use your idea, you would have to close the control loop with the MLT temperature, as with a significant difference in temps it is unlikely that you will be able to correctly guess at the temperature delta that will work. That is an inherently tricky concept though.

As an FYI, I have a doctorate in mechanical engineering, specifically in controls. If I seem long winded, pedantic, and overly confident of my own correctness, its because I am

The physical link between the HLT and MLT is only as strong as the heat exchange between them. If you have weak heat exchange (low flow and a short coil), it takes a long time to transfer energy to the MLT from the HLT.

Now there is a PID controller that is taking the MLT temperature tracking error in, and calculating the voltage output to the heating element in the HLT. When the MLT is slow to respond to the controller output, the controller will put out EVEN MORE to the heating element.

Pretty quickly, the heating element will get railed at full power. And at some point, the MLT temp will rise and get to the set point. Once this happens, you will probably have a very hot, if not boiling, HLT. You can't really actively cool the HLT (only heat. we have a one-sided control action), so it will probably cause the MLT to get much warmer than you intended. Eventually (hours?) you will get to a nice steady-state condition, and everything will be great. Before that happens, you will have taken your mash and HLT through big temperature sweeps.

The control solution, at least with a PID, is to decrease the I significantly, and maybe even back off on the P (you can't expect more control performance than is physically possible, after all). This means that your response will be very slow, and probably have significant steady-state error.

Luckily there is a solution

Since you already have 2 pumps, you can close 2 control loops. One is the HLT. Set it at something pretty high. Maybe 180-190 or so, but that is a guess. So, the HLT maintains its own temps by controlling the heating element while constantly recirculating (the faster the better). Close another loop around the MLT temp. This time, I'd use a hysteresis control law (unless you really want to mess with variable speed pumping) to turn the pump on/off. When the MLT temp drops below a certain point, the pump turns on. When the temp rises above a different point, the pump turns off. This should give, more or less, a set temperature range that the MLT will stay within. I'd shoot for maybe a 2-3 deg dead-band between the two set temps. Enough that the pump isn't cycling every few seconds, but not so much that you can't control the enzymatic reaction.

Wooboy, that was a long one.

as an FYI, that is how most glycol systems work. You maintain a nice cold glycol bath, and recirculate as needed to cool various components. That way, you can account for losses, have different components at different temperature, etc... check out the bad-ass glycol chilled plastic conical thread for a well-built example.


I'd be careful about a too hot HLT though, when it comes time to sparge.

Thank you for the input! With you're expertise, I'd like to ask a couple more questions, if you don't mind:

In EE, we use PID controls for sensitive systems. Ramping voltage, creating waveforms, etc. However, In a MLT, I don't see the need for such sensitivity. There is so much thermal mass, and assuming a +/-1 degree temperature swing is acceptable, would it not be acceptable to follow these steps:

1) Heat HLT to +3/4 degrees hotter than the desired mash point.
2) Given a thermometer reading at the end of the HEX, a constant pump velocity, slightly turbulent flow is assumed constant within the coil, water is at rest outside of coil at specified temp, can we not assume that the Mash Temp will follow the output of the HEX?
3) Should the temperature drop, we simply heat the HLT another degree or two, and see if there's a change.

With all the fudge factors we have to assume with a HEX system, what is the point of developing a true PID control loop? When you designed yours did you not have to run multiple tests before you began to attain an acceptable temp control?

So my question: Is a PID truely necessary if we have monitor the temperature of the recirculation mash as it exits the HEX?

Based on 1/2" tubing, copper, 25', a maximum of 1.5' height difference between mash water level and recirculation input, the pressure loss is 1.6psi or 44" water column, adding height difference I see 62" water column. The USSolarPumps are rated for 9.8' water column at 3GPM.

I estimate I can still achieve 1GPM at full run, including line loss from silicon tubing and the 2 couples leading out and into the mash tun.

I'd be careful about assuming you can calculate the head loss analytically. The equations out there that 'work' for coiled tubes, such as Dean's equation, are all experimentally determined, and may not directly apply without more fudge factors. Add in the effects of various fittings, 90 deg turns, wort viscosity (1st runnings are pretty sticky), and I'd think it would be tough to nail down a number with any confidence.

I've never used one of USSolarPumps, so I can't say with any confidence... I have my own experience with my chugger pumps and my coil to say that my flow sucks. I have minimal fittings, all TC connections, so its just the coil holding me back. Or karma. You'll have a less restrictive flow with a weaker pump, so who knows. I'm curious to see the answer though.

About the heat exchange idea, I drew up this block diagram for the discussion. I was making block diagrams anyways for work, and it helps to have pictures. It assumes constant HLT temp. 1/s represents an integration.

Essentially, the thermal energy you are adding (or subtracting) is the input from the HEX minus the losses to ambient. Depending on the 2 different coefficients, as well as the deltas between HLT and MLT and Ambient, there will be some steady-state MLT temperature. Even if you assume that the HEX effluence matches the HLT in temp (with your coil and proposed flow rate, it should be close at least), that doesn't mean that the MLT will match the HEX effluence unless you have a very high heat transfer coefficient (i.e., high flow rate). And if the coil effluence matches the HLT temp, the only way to up the heat transfer coefficient is to increase flow rate.

Given the block diagram I supplied, and assuming constant heat transfer coefficients, thermal masses, and HLT/Ambient temperatures, you can conclude that the steady-state MLT temp will be somewhere between the HLT and Ambient temps. The thermal mass, 'C', just dictates how long it will take to get there.

Without something closing the loop around a mash temp (MLT exit, recirc input, probe in the middle), there is no way of directly maintaining mash temperature. You can try and calculate what HLT temp would give you the correct MLT temp, but that calculation will not be very accurate and you will end up adjusting it a lot. Essentially, closing the loop with YOU as the controller.

If you look around the forum at the various HERMS setups, you'll see a few trends. Many folks will insulate the MLT, which pushes the ratio of h_hex/h_amb in your favor. Or, they'll switch to larger coils (like I need to). Or, they'll set the HLT a few degs higher to make up for the difference.

If you look on Kal's website, he doesn't do any of those things. I would guess that with his 1/2" coil and a center inlet march pump, he is pumping fast enough that the MLT tracks the HLT within a deg.

I'd say, spend the money on a nice wort pump. The center inlet chugger can be had for 140 if you shop around, sometimes even less during a sale. They make a little noise and cost more, but you'll head off potential problems down the road. If a cheaper pump worked just as well for HERMS, you can bet that a lot of homebrewers on the forum would use them. As it is, most use march/chugger.

Plus, if this was your plan, I can get a crazy whirlpool going with my center inlet chugger. Not sure a USSolar pump can do that.

I didn't mention it, but this is a hell of an undertaking for a senior design project. Back in 2006, mine was to design a machine to roll sand paper into belts for 3M. Your project sounds like a lot more fun.

Im a senior at Purdue CoT in Kokomo, also making brewing system. Although we aren't in teams. For cost purposes I'm only automating the mash. I currently have my system set so the HLT maintains a set temp x degrees higher than the set mash temp, and the mash temp is controlled by turning the pump on and off recirculating through the hex.

Boiler Up!!

A few old android cell phones make for cheap and easy to setup Webcams. Always a nice thing to be able to see things remotely when using an automated process.

Sorry if it's sloppy, this was sent from my phone

KuntzBrewing said:

Im a senior at Purdue CoT in Kokomo, also making brewing system. Although we aren't in teams. For cost purposes I'm only automating the mash. I currently have my system set so the HLT maintains a set temp x degrees higher than the set mash temp, and the mash temp is controlled by turning the pump on and off recirculating through the hex.

Boiler Up!!

Click to expand...

Phew, I was starting to think they don't teach KISS is engineering schools anymore.

Spintab said:

Phew, I was starting to think they don't teach KISS is engineering schools anymore.

Click to expand...

Behind every simple design is thousands of hours of design simplifying the interface and circuitry

Sent from my DROID RAZR HD using Home Brew mobile app

Question: had you ever considered RIMS rather than HERMS. Nothwithstanding the fact that wars have been fought over lesser debates, from a project standpoint (programming, control loops, pumps) I would think it would be easier to implement and control. Just my tuppence.