I think you have proved what I have been told and actually found works in practice over the years (and have actually said on here, but had some sceptical response), the water level in the boiler affects the shot temperature. The interesting thing is you have also proved it affects shot stability

Low water=higher shot temps (but poorer steaming)
High Water levels = lower shot temps, but better steaming (unless it's far too high)

All for the same given pressure (or temperature).....the reason as I understand it is because of the HX orientation within the boiler, passing through both water and steam. Steam is apparently 4 times more efficient at transferring heat than water. This means your brew temperatures are up overall with low water levels(after a cooling flush). You could confirm this by doing the same test with NO 50ml cooling flush and you should find the max shot temperatures would be the same whatever the boiler fill level as the water sitting in the HX should be exactly the same in all cases.

This efficiency of steam to transfer heat is why the lines also group together more on the lower water levels. I believe many years ago I read about some manufacturers passing the HX only through steam in an attempt to improve shot stability and also make the HX more efficient for shot after shot... and I am sure they are propbably still doing this today. I wish I could find the study/paper from the manufacturer as it makes interesting reading.

Nice bit of work and interesting reading...thanks

DaveC wrote:Steam is apparently 4 times more efficient at transferring heat than water.

While I am not an engineer, this seems as though it inverts the laws of physics. Is not a body of water of enormously greater efficiency at transfer of heat than a body of gas (steam) at the same temperature?

DaveC wrote:I think you have proved what I have been told and actually found works in practice over the years (and have actually said on here, but had some sceptical response), the water level in the boiler affects the shot temperature. The interesting thing is you have also proved it affects shot stability

I would be reluctant, at this stage, to say that I "proved" anything. Rather, I'd put this into the category of "interesting observation, deserves further study."

I am certain at this point that boiler water fill level has an impact on espresso shot temperatures, especially from the Cimbali Junior S Pourover espresso machine I tested. I would like to confirm this observation in my Rotary pump machine which has autofill. After adjusting the autofill probe down to reduce the water level to the minimum, there are still about 21 oz of water that come out through the water wand when I drain the boiler. At this fill level, I get good but not great shot temperature consistency, however there is without doubt quite a bit more water in the boiler than on the older machine with manual boiler fill, at the "minimum" level on the sight glass.

I have a spare generic autofill probe, and am going to do some experimentation the next time I drain the boiler on the rotary (probably this upcoming weekend). If I can safely get the water level down significantly further, by 10 or more ounces and hence mirroring the situation in the old Vibe machine, then I should be able to test the theory with some new shot temperature curves. Of course, it would be nice to have some other people confirm or disprove my observations, on other equipment, which would give us a better idea of how generalizable these observations are.

ken

Richard wrote:While I am not an engineer, this seems as though it inverts the laws of physics. Is not a body of water of enormously greater efficiency at transfer of heat than a body of gas (steam) at the same temperature?

Well not really when you consider weight for weight and the latent heat of vaporisation/evaporation (which is pretty enormous as water is a dipole and really doesn't like being split at the molecular level....all those little magnets). When hot steam comes in contact with a cooler surface, it condenses. The change of state from a gas to a liquid releases much more energy than what would be released by just cooling the same weight of water by one degree.

In reverse it's also the reason why evaporating sweat is so much better at cooling you than completely wet skin.


http://en.wikipedia.org/wiki/Heat_pipe

"In summary: inside a heat pipe, "hot" vapor flows in one direction, condenses to the liquid phase, and migrates back in the other direction to evaporate again and repeat the cycle. One reason for the effectiveness of heat pipes is the amount of heat that an evaporating fluid absorbs and then returns when it condenses. For water for instance, to evaporate one gram of water takes as much heat as would be needed to raise the temperature of that same gram of water by 540 °C."

Or possibly a more relevant example as similar temperatures and pressures are used.

http://missvickie.com/workshop/howdoesit.html

"Steam has six times the heat potential when it condenses on a cool food product. This increased heat transfer potential is why steam is such an effective cooking medium."

The HX pipe with it's cold water inside can be considered a cool food product in tranferring the example to a coffee machine.

Oh Ken the proof I refer to is only the relationship of water level to brew temperature....which I was told about years ago, found actually works in practice (as you have). But because it seems counter intuituve to lower water level and increase brew temps and vice versa, seems to be contovercial.

First and foremost, thanks Ken for taking the time to post this data although I'm teetering a little on the validity of your conclusion. My rationale being that if I remove one of the outliers from one or more of your four graphs representing mid and high level shot temperatures, shot consistency would be fairly uniform.

What is, however, easier to see is the decrease in average shot temperature as the boiler water level is raised or the increase as the level is lowered. As DaveC points out, he brought this subject up in another post and was met by some skepticism, certainly by me and perhaps others. The skepticism was not well founded.

The overall Heat Transfer Coefficient (U) for a Steam-Copper-Water "system" is 205 Btu/ft2-hr-F or 1160 W/m2-K whereas the overall Heat Transfer Coefficient (U) for a Water-Copper-Water "system" is 60 to 80 Btu/ft2-hr-F or 340-455 W/m2-K

The above data was extracted from this reference: http://www.engineeringtoolbox.com/overa ... d_284.html

As the water level in the boiler is lowered, the surface area of the hx exposed to steam is obviously increased. As the overall heat transfer coefficient (U) for a steam-copper-water "system" is about three times that of a water-copper-water "system", a greater amount of heat is transferred (Btu's per unit time) to the water inside the heat exchanger. This increases the temperature of the water inside the heat exchanger in accordance with established equations of heat transfer.

This hotter water, in turn, increases the average temperature of the grouphead which is the dominant driving factor in shot temperature changes.

There are two problems I have observed in shot temperature consistency which I attribute to higher than needed boiler fill. The first is "outliers," and I have scads of shot series and walk up shots I've made whose graphs have never been posted (and I've posted tons of graphs over the last year and a half or two). Outliers are one major problem that I'd like to eliminate, and are not infrequent, rather they are common with the test runs I've done. And the test runs I've done before have been done at higher boiler fill levels than I'm testing now, with early promising results that seem to reduce or eliminate the outliers, especially the worst of them.

The 2nd problem is with consecutive shot series. With high levels of boiler fill, the element simply can't heat the water surrounding the heat exchanger fast enough to keep up, and shot temperatures decline the further one goes into shot series. With lower boiler fill, my hope is that the reduced energy requirements of heating steam vs. water will make it easier to provide the heat, via steam, to the HX to allow shot temps to be maintained.

All of this requires further testing, however, before I can be certain. Early results are promising.

ken

erics wrote:First and foremost, thanks Ken for taking the time to post this data although I'm teetering a little on the validity of your conclusion. My rationale being that if I remove one of the outliers from one or more of your four graphs representing mid and high level shot temperatures, shot consistency would be fairly uniform.

What is, however, easier to see is the decrease in average shot temperature as the boiler water level is raised or the increase as the level is lowered. As DaveC points out, he brought this subject up in another post and was met by some skepticism, certainly by me and perhaps others. The skepticism was not well founded.

The overall Heat Transfer Coefficient (U) for a Steam-Copper-Water "system" is 205 Btu/ft2-hr-F or 1160 W/m2-K whereas the overall Heat Transfer Coefficient (U) for a Water-Copper-Water "system" is 60 to 80 Btu/ft2-hr-F or 340-455 W/m2-K

The above data was extracted from this reference: http://www.engineeringtoolbox.com/overa ... d_284.html

As the water level in the boiler is lowered, the surface area of the hx exposed to steam is obviously increased. As the overall heat transfer coefficient (U) for a steam-copper-water "system" is about three times that of a water-copper-water "system", a greater amount of heat is transferred (Btu's per unit time) to the water inside the heat exchanger. This increases the temperature of the water inside the heat exchanger in accordance with established equations of heat transfer.

This hotter water, in turn, increases the average temperature of the grouphead which is the dominant driving factor in shot temperature changes.

Hi Eric:

Are you sure that vapor phase heat transfer coeficients are higher than for liquid phase. That doesn't seem right to me. Are you sure that the values you quote are for liquid and vapor phase at the same temperature?

-Greg

No, I'm not 100% certain - I was simply quoting from the reference given. Now that I've donated 6 boxes of books to the UMd engineering library, you ask me this?
However, I think I know where to look - http://www.spiraxsarco.com/resources/st ... orials.asp

As you know, the "search" for H (convective heat transfer coefficient) and eventually U (overall heat transfer coefficient) is the heat transfer garu's dream come true.

I spent some time yesterday monkeying with the autofill probe in my plumbed in Cimbali Rotary Junior D, which (unfortunately) has boiler autofill. Did I ever tell you before that I HATE autofill?

I learned a number of things including that it is hard to get the autofill set at a lower level than a certain point, because the probe simply can't make the curve into the boiler when bent beyond a certain angle. I could experiment with more rounded curves, but since these probes should be cleaned occasionally (even soft water will put white powder on the probe which interferes with conductivity) so you do need to be able to remove and insert them, and whatever you do should be repeatable, unless you enjoy doing shot curves with a Scace device. Finally, the autofill doesn't set an exact level, rather it sets a range. Depending on the level of the fill asked for, at least in my machine's boiler, that range has a "slop factor"of as little as 3 and as much as 5 oz. (roughly 90 to 150ml).

So, there are limits to what you can do with autofill systems, unless you want to alter them in a way that allows you to turn on and off the autofill. I believe Cannonfodder has done that with one machine. Alternatively, the poor man's way of doing this would be to drain a few oz. of water out the water wand each day, and most probably you will go back toward the lower level of the fill range.

I have now reduced the amount of water that will discharge through the water wand from a prior range of 21-26oz, to the current 16 to 19 oz. This is an obvious reduction but maybe not a huge one, but the shot temperature curves I have obtained with the lower fill level are tighter and more repeatable than before.

I will use a set of curves as a sort of "before and after" illustration, with "Random Walk-up Shots." Basically, as I define it, "random walk-up shots" are shots preceded by a standardized flush (50ml in the case of my machines) and measured at intervals of 10 minutes or longer in between shots. As a general practice in actual use of my machines, when my machines have been idling for long periods of time, say exceeding 4 hours, I generally do one of these 50 ml flushes and pull a blank shot, then wait a few minutes before pulling an actual shot, since the first shot from a long idle period tends to be too hot. I have done this when I have generated curves such as these also.

Here, on the rotary machine, are "before and after" random shot series at 232F boiler temperature. As with earlier observations on the (more easily adjusted) vibe machine, the shot temperature changes with boiler fill levels at the same temperature. Also note that the labeling of these graphs is a little confusing; the top one reflects a boiler fill level with the unmodified boiler probe set to the lowest point, and the bottom graph shows what happened after I bent the autofill probe as far as I could do so while still getting it inside the boiler, so there was a reduction in fill level from the top curves to the bottom ones:




vs.



I have other sets of curves that illustrate the same things, e.g. that the observed shot temperature seems to go up, and the curves become more repeatable and in a tighter range, as the boiler fill level is reduced. I will probably do a little more experimentation with this, to see if I can get the fill level even lower, although I believe that these curves demonstrate the basic point is valid in autofill systems also.

People who are seeking tight shot temperature control, who are using electronic temperature control of their boiler as a way of achieving this, should experiment with boiler fill levels. The shot temperatures you get are going to be related to the fill level, so even if you don't intentionally change the fill level, you should try to use the SAME fill level consistently. In a manual fill system this is easily done. With autofill it is harder, and may include regularly draining some water out of the boiler in order to get the autofill to reset the level at a consistent, probably lower, point.

ken

There seem to be two effects on both machines. The first is to reduce or eliminate "outlier" shots. Presumably, this is because lower water mass reduces overshoot, or gets the boiler back to its base state more quickly. The second effect is the head scratcher: your lower boiler fill curves have a flatter temperature profile.