Deciding where to put a temperature sensor in a HERMS ("Heat Exchanged Recirculating Mash System") can be perplexing, especially if the sensor is feeding back to a "PID" controller that automatically manages the heating element in the HLT ("Hot Liquor Tank"). And therein lies the source of disagreements. HERMS uses the heat exchanger to manage the heat in the mash liquid, and therefore the mashtun: But HERMS does not control the flow of heating liquid through the heat exchanger (as is the case in many other process control examples), instead it controls the temperature of the water in the HLT. This difference opens up different ways of thinking compared to many examples published about heat exchanger process control. RIMS heats the mash (process) liquid directly so avoids these arguments (but generates a few of its own).
Let's start with a somewhat stylised diagram of a HERMS setup.
RIMS ("Recirculating Infusion Mash System") is much the same but the element ("E") is actually in the recirculating mash so there is no HLT holding the heat-exchange coil of recirculating mash. The element is going to be electric in RIMS but a gas burner could be used to heat the HLT for HERMS (more likely it is electric too - I've never seen a gas version but they exist).
The diagrams illustrate the four likely locations for placing a controlling temperature probe: A, B, C and D. The pump ("P") recirculates the mash. Ideally all four locations read the same temperature, and they will for all minor changes to temperature; the actual location of the temperature probe becomes important when big hikes in mash temperature are being made, either when initially heating to the mash temperature or stepping the mash temperature up for stepped mashes or "mash out" at the end.
The advance that has come with both systems (HERMS and RIMS) is the "recirculating" bit. The mash liquid is separated from the grain (by a false bottom or the like), heated, and then returned to the grain. Previously insulation was the prime choice for keeping the mash temperature steady, or else heat was applied directly (there are also methods of manually separating portions of mash and heating it, such as in "decoction" mashing). The problem with directly heating the mash is the high proportion of solids in a mash which limits heat transfers to slow conduction, not faster convection (with associated currents), which will create pockets of cooler or hotter mash unless stirred well, and the solids can burn to the heating surfaces. Constant recirculation also stands in for convection currents, keeping the temperature throughout the mash more consistent.
"A" is one logical location for the temperature probe controlling heating of the mash i.e. in the tank with the heat source (the HLT). But is it the best? The most likely controller being used is of the "PID" type, because they are very good, but unless they have been very carefully tuned they can also be horribly slow at converging the set (desired) temperature with the actual temperature. And placing the probe at "A" has the effect of exaggerating this slow convergence. Tuning could accept a small temperature overshoot to speed up convergence, but most users are not going to be developing the necessary knowledge to fiddle about with these controllers in such detail. As a location for the controlling temperature probe in HERMS "A" may require a complete shift of thinking to select. No wonder the sensor location for HERMS causes such division of opinions.
"B" is a logical location for the temperature probe in RIMS because there will be no location "A". Location "B" (and "C" and "D") is actually in the recirculating mash stream. Many documented examples of heat-exchanger processes will show the location of temperature sensors at "B".
"C" as a location seems to be ideal because it's in the mashtun (MT) with the mash. But even with recirculation, temperature through the thick stodgy mash can't be guaranteed uniform so isn't a good candidate for a sensor controlling mash temperature.
"D" would seem a reasonable location, but there would be considerable lag between heating and seeing the result at "D". Meanwhile there would be undesirable over-heating of the recirculated mash unless very careful tuning was undertaken (far more so than if "A" was used - somewhat defeating any point of not using "A"). "D" is a good location for monitoring because it will help confirm when the system has achieved equilibrium (i.e. heating has done its work to reach the desired temperature and the system is now just maintaining it): Without such monitoring the system can be the source of uncertainty.
So, what of "B", the same location as would be logical for RIMS (the HLT and HERMS coil sort of replaces the RIMS tube). Can this be a better location than "A" for HERMS? When temperature changes are small the sensors at "A" and "B" will most likely be reading the same temperature. But for big jumps the heat exchanger may start lagging behind the speed at which the HLT is being heated. Meanwhile if the sensor is at "B" the controller could be allowing the HLT to over-heat (the controller is being told the temperature is too low): But that will also mean the heat-exchanger will not be lagging behind the desired temperature by quite so much (because more heat is being provided to the heat-exchanger). And as long as the mash is sucking enough heat out of the HLT for the over-heating to stay within "reasonable" limits, the over-heating should get less and less as the desired temperature is converged on.
So "B" is ideal? Minimal tuning yet much of "A's" slow convergence is countered. This does depend on the dynamics of the brewing setup design but this can be ignored for most brewery designs. Some may not want to risk "B" and will stick with the "safe" sensor location of "A" (over-heating the HTL is very unlikely). Even if a brewing setup is found to be incompatible with this "B" approach, a far coarser and cheaper controller (than a PID controller) could be used on the HLT to limit its potential over-heating. The amount of lag between the HERMS coil and the HLT will depend on the flow rate of mash fluid through the coil (and the efficiency of the coil - note "rate" must take in the possibility of a stuck mash) so using location "B" releases the convergence of actual MT temperature and desired MT temperature from those factors too.
But there is a snag (for HERMS). If any sensor location other than "A" is used to control the temperature of the MT, the HLT cannot be heated using the same PID controller unless the recirculation system is operating. This might not be an issue for many, and it will not be a problem to work-around for the handy folk who would like the challenge. But the limitation may prove unsurmountable for some, in which case using location "A" is the only option. Another snag when not using "A" (for HERMS) is if recirculation is interrupted (e.g. stuck mash): The controller may continue to instruct the HLT to keep heating (until it is boiling unless some sort of fail safe is worked in).
So to gain the potential benefits of using location "B" for HERMS (i.e. possibly faster temperature ramps) you must accept the HLT is out-of-bounds for ad-hoc little heating jobs when the recirculation isn't running. But would the HLT ever be required when the recirculation isn't running? The handy might develop a scheme to switch between locations, or work in a second controller for independently running the HLT, but such complexity is outside the scope of this article. But this division of options should keep arguments raging for some time to come. Or avoid this divisive option and plump for RIMS, but be warned the choice of RIMS over HERMS is just a source of even more arguments!
Finally there are the "one-pot" designs such as the popular "Grainfather":
Look at it carefully. All that has changed from the RIMS diagram is the position of the element. The "One-Pot" designs are RIMS! The temperature probe will frequently be sited at position "D", the position not recommended in the discussion above. But position D and B are near enough identical in this case. There is no position "C" because this design will have the grain contained in a basket, or "malt pipe", which makes position "C" awkward if not impossible.
I got this for big transfers; 
From what I gather, no method is perfect unless you’re a genius with circuits and wiring and have the ability to factor in readings from multiple points in the brewing rig. 
