While I was doing a 15 gallon turbid mash with 5 temperature steps, I decided to do some research into performance of different temperature control algorithms.
Common practise on a HERMS system is to control the mash temperature indirectly by varying the temperature of a direct-fired hot liquor tank. Wort flows out of the mashtun, via a pump, and through a heat-exchanger coil in the hot liquor tank and then returns to the mashtun. If you want your mash to be 65℃, then setting your HLT to 65℃ is a reasonable approach. In a world with no thermal losses, this model should work well.
In reality, thermal losses from the HLT, hoses, and pumps leads to a condition where there is a small differential between the HLT and mashtun. My home brewhouse, which consists of an 80 litre (20 gallon) mashtun, a 60 litre (15 gallon) HLT fired by a 5500W electric element, and a 14m (45 foot) HERMS coil driven by a pump that moves 10 litres of wort per minute through the coil typically requires the HLT to be set around 1 - 1.5℃ hotter than the mashtun when brewing 60L batches. This can vary slightly based on the temperature setpoint, mass of grain, volume of water in the mashtun, and volume of water in the HLT.
Here is a plot of my system's response when ramping the mashtun
from 50℃ to 65℃.
The blue curve represents the temperature in the HLT. The green
curve represents the temeprature of the wort at the exit point
from the heat exchanger coil where it returns to the top of the mashtun.
The red curve represents the temperature of the wort at the
bottom of the mashtun as it exits the vessel and heads towards
the pump. The thermal gradient in the HLT is minimized by a
recirculating pump which keeps the tank stirred and fairly
uniform. You can see here that my HLT setpoint is just over
66℃. There is a control deadband of about 0.5℃ around
the setpoint, which is why a zig-zag response can be seen.
If an equal mass of hot and cold water are mixed together, the resulting final temperature of the combined system is a weighted mid-point between the two initial temperatures. e.g., 10L at 20℃ and 10L at 30℃ yields 20L at 25℃. In a heat-exchanger like a HERMS coil, the volumes don't actually mix, but instead flow past each other at a specific rate for a specific period of time. This results in a slight cooling of the water in the HLT and a warming of the wort flowing through the coil. The rate at which heat transfers across the stainless steel wall of the heat exchanger coil is determined by its thermal conductivity and the difference in temperature on the inside and outide. In a nutshell, the larger the difference in temperature on either side of the wall, the quicker the transfer of energy (heat). This is the same reason an ice cube melts faster in hot water than in cool water. A final point to note here is that the temperature of the wort leaving the heat exchanger will never fully reach the same temperature as the water in the HLT, since the system is trying to reach a point in between the temperature on the inside and outside of the coil. The wort can get closer and closer to the HLT temperature the longer it stays in the coil, but it will only asymptotically approach the HLT temperature. You can see this in the above graph -- when the HLT and wort temperatures are further apart, the rate at which heat is transferred is faster (steeper slope), but when the temperatures are closer together (at the start and end of the ramp), the rate of change is slower
By allowing the temperature of the HLT to overshoot the target mash temperature by a calculated amount, it is possible to reduce the time required for the mash to reach it's final temperature. This is because the temperature differential in the heat exchanger is being increased, thus increasing the rate of heat transfer.
You'll note that 2 new curves are plotted here -- T and G. The T curve indicates the overall system temperature, which is the weighted sum of the thermal masses of both the HLT water and the mash. G is the final weighted goal state for T.
The formula I am using to compute the current system temperature
(T) is:
You can see that the HLT reaches 66℃ at about the same time as in the previous case, but instead of cutting the power at that point, the HLT is allowed to continue heating up until the overall system temperature (T) reaches the goal (G). Again, the reasoning here is that the heat-exchanger will finish the work of moving the excess energy from the HLT to the mash. As a result of this HLT overshoot, the temperature difference between HLT and mash is allowed to remain larger for a longer period of time, thereby increasing the rate at which energy is transferred to the wort. Additionally, without any further energy input, the overall system temperature (T) cannot increase any further, which ensures the mash doesn't overshoot. When the 2 graphs are co-plotted, you can see that there is a time savings of around 25%, or over 8.5 minutes:
You can see the simple control algorithm's impact on the mashtun temperature as the blue MLT-A curve, and the overshoot algorithm's impact as the red MLT-B curve.
As a note, my turbid mash schedule required a dough-in at 45℃ and rests at 50℃, 65℃, 70℃, and a mash-out to 76℃. In the end, for this 15 gallon batch, I saved almost 30 minutes by using this overshooting control algorithm.
Is it possible that the over-heated HLT could lead to de-naturing of the enzymes in the wort? If you are ramping temperature for dough-in or mash-out, this shouldn't be a concern, however if you are doing temperature ramps for alpha- or beta-amylase rests, then it might be. Let's take a quick look at a section of one of my temperature plots for a 65℃ - 70℃ ramp.
You can see that the green curve does exceed the mash setpoint of 70℃, but recall that the green curve in my system is always slightly higher than the mash temperature, due to the nature of the losses in the system. The difference in temperature from my setpoint appears to only be around 0.5 - 0.7℃ in this case. While that may be enough to impart an accelerated denaturing, keep in mind that the wort will only reach this peak temperature just as it is finishing it's run through the HERMS coil. If the wort is only overheated for a brief moment before rejoining the full volume of the mashtun and equalizing, then the denaturing effect is greatly reduced.
Here are a series of comparative plots for my 40kg setup ramping between different temperature points and a note on how much time was saved using the overshoot algorithm. The 40kg figure comes from the specific grain bill I was using for my last brew which was approximately 34kg of water plus 13kg of grain (34 + 0.45 × 13 = 39.85kg equivalent thermal mass). My HLT was filled with 50kg of water.
After some discussion with members of my local club, some concerns were raised that the performance of my system is different from that of others. Specifically, the temperature of my mashtun was lagging by several degrees when ramping, wheras others reported only 1-2 degrees. It's true. Different behaviours will be seen when system variables are adjusted:
Some other musings gathered during the investigation:
In certain system configurations, it can be beneficial to purposely over-ramp the HLT in a HERMS brewing setup to reach different temperature steps if:
I am considering leveraging the knowledge gained from this experiment over to the STC1000+ project for use in fermentation chamber temperature control.