Practical Advice On Managing Wine Alcohol Levels

Fourteen percent, more or lessThe other day I came across this fact sheet: “Reducing Alcohol Levels In Wine” published by the Australian Wine Research Institute (AWRI). Directed at the professional winegrower, this is the best agenda-free piece on wine alcohol levels I have read, period. It’s worth the interested reader’s time.

All of my own efforts to manage alcohol levels in our wines are mentioned here. In the vineyard every year we reduce canopy leaf area to balance crop load, and I have found that irrigating to 85% of evapotranspiration demand right up to harvest prevents runaway sugar accumulation. I have always disdained wines with over-ripe flavors, and so have always picked at the earliest date that I find the various components of the grape to be quote-unquote “ripe” – a personal definition, but one that I am happy with.

I found it amusing that the AWRI paper discusses water adds under the heading of “blending.” Adding a “reasonable” amount of water, for one reason or another, is a common practice in winemaking. We just don’t talk much about it.

I was left scratching my head over the mention of glucose oxidase to decrease the level of fermentable sugar in juice or must. I recall reading a few research papers in the 1990s about this, but didn’t think the technology ever made it out of the lab. I honestly don’t know of any winery that uses this enzyme. Nor have I ever come across a commercial preparation for use in wine. So, pace, “natural” wine aficionados.

Fermenter design does make a difference. I prefer to use fermenters with a must depth of 38″ during peak fermentation, regardless of diameter, and seek to achieve peak fermentation temperatures of around 90° F for my red wines. I have empirical evidence that this approach reduces our so-called “conversion ratio” (the percent alcohol immediately after fermentation divided by the Brix before fermentation) by up to 5%.

By contrast, I have found no consistent evidence that yeast selection has any effect on alcohol level. Whether I conduct a ferment without inoculation, or by inoculation with a selected commercial strain, the final alcohol is the same within measurement error. Incidentally, these days I start every fermentation without inoculation. If the initial Brix is high or if the ferment shows evidence of stress, I inoculate with a commercial strain I feel most suited to the variety. In effect, all our ferments are conducted by mixed strains of yeast.

The AWRI paper discusses the most obvious, the most used, and the most discussed (and often reviled) method of alcohol level management: physical removal of alcohol from finished wine by reverse osmosis or vacuum distillation. I have experimented with these methods on a limited basis with mixed—mostly negative—results. My biggest concern with large-scale alcohol removal is that the wine is nearly always rendered “hotter” by the treatment. I speculate that this is due to removal of ethanol at a faster rate than alcohols of three carbons or more by the processes.

The article mentions de-alcoholizing small parcels of wine and blending back. I have had some good results with this approach and I am experimenting with this method on an ongoing basis, because of the next topic discussed in the article: loss of alcohol by evaporation during barrel aging.

In fact, during barrel aging in our cellar the alcohol level of the wine increases by up to 1.2%-1.5% over two years. During barrel aging, the wood of the barrel acts as a semi-permeable membrane. Wine components inside the barrel migrate through the wood at various rates and evaporate from the outside surface. My a priori assumption is that the rates of migration of water and alcohol are dependent on the differences in concentrations between the inside and outside of the barrel.

Let’s say I put a wine to barrel at 13% ABV; this wine is approximately 87% water. In our barrel cellar, the concentration of alcohol in the air is essentially 0%, while the relative humidity averages about 35%. Water leaves the barrel faster than alcohol because 87%-35%=52% is four times greater than 13%-0%=13% (52/13=4); therefore, the thermodynamic drive for water to leave the barrels is 4x the impetus for alcohol to escape.

The AWRI paper discusses how alcohol levels decrease over time when the average relative humidity of the barrel cellar is 70%-90%, but also discusses the negative issue of mold growth in the cellar in this wet environment. Our barrel aging area was not designed to be wet, and we also store cased goods in proximity to our barrels. Humidification of our cellar is not an option.

My intent is to experiment with vacuum distillation of the wine I use to top our barrels. If we decrease the alcohol level of the topping wine, I believe we can slow the rate of alcohol increase in our barrels over time in our dry cellar environment.

14 thoughts on “Practical Advice On Managing Wine Alcohol Levels

  1. SUAMW

    Hi John,

    Please elaborate on this:

    ” seek to achieve peak fermentation temperatures of around 90° ” – do you mean to say that you a) intentionally/actively encourage your fermet temps to go up to 90, or b) take intentional/active measures to lower your ferment temps below 90?

    How does such a ferment temp affect aromatics? Is it not the conventional wisdom that higher temps blow off aromatics?

    As to water and alcohol migration rates in oak: I would expect the two to be different (one is an organic solvent, the other is not – and the interactions with oak might be different) band to be affected by toast as well. Thoughts?

    1. John M. Kelly Post author

      The must depth of 38″ allows Pinot ferments to achieve at least 90° F peak fermentation temperature. I use frequency of punchdowns at peak activity to try to limit the temperature peak to 98° F, as the yeast start to really stress at higher temperatures. I don’t use any active control to raise or lower temperature.

      Fermentation at any temperature blows off aromatics. I’m completely unconcerned about this effect in the temperature range I work in, as my finished wines are not “deficient” in any way that matters to me. I am much more concerned about “off” aromas generated by stressing the yeast.

      I have considered the effects of solvent polarity on the rate of migration through wood, but I am certain that this second-order effect is quite small relative to the psi generated by the concentration differentials.

      Toast might have an even smaller effect through damage to the wood structure, but I have never noticed any difference between the medium toast and heavy toast barrels I use. That said, consideration of rate of loss would never enter into my calculus of what mix of toast level to use. I may have said it elsewhere, but I pay no attention to toast level at all—I simply ask my coopers to give me their favorite toast based on the long experience they have with barrel making.

  2. Kevin

    Hi John,

    Thanks for taking the time to write this and for sharing your experiences and your approach. I was particularly interested in your observations about increases in alcohol during barrel aging. I did some quick looking around and found a few interesting articles.

    Mosedale in 1995 has an article titled ‘Effects of oak wood on the maturation of alcoholic beverages with particular reference to whisky’ (in the journal ‘Forestry’) — he makes a few observations:

    “[P]ermeability also varies between different liquids. Kiseleva and Zoldners (1986) found that pure water diffused through birch wood four times faster than pure alcohol, when restricted to passing through cell walls, as is likely to be the case when movement through vessels is blocked by tyloses.”


    “Guymon and Crowell (1970) described how humidity affects the final proof of maturing brandy. Low humidity produces an increase in alcohol content due to the evaporation rate of water being higher than that of alcohol. In high humidity conditions, the proof declines, as more alcohol than water evaporates (alcohol evaporation being independent of humidity). Evaporation may then result in further effects, due to the changes in the proof of the maturing distillate.”

    Kiseleva, T.B. and Zoldners, Yu.A. 1986 Kinetics of penetration of phenolic alcohol solutions into wood cell walls. Khim. Drevesiny 6, 58-64.

    Guymon, J.F. and Growell, E.A. 1970 Brandy ageing: some comparisons of American and French oak cooperage. Wines and Vines 51, 23-25.

    thanks again for writing and sharing these technical posts,

    1. John M. Kelly Post author

      Thanks, Kevin! We only use French oak (Quercus robur and Q. petraea) that have a very low level of tyloses—this is why the staves must be split rather than sawn. I have made observations of across-the-grain wet-ability of French oak chips with water and with 190° proof grape spirits, and never noted large differences, but this was not any sort of experimental setup. I’m not sure of the relevance of the results observed with birchwood, and it is not clear to me if the discussion involves diffusion with the grain. I’d be interested if you could point me to an online English translation of the Kiselva, Zoldners & Yu paper. I’ve got the Guymon & Growell paper in my files.

  3. Bob

    Hi John,

    There is a good reason that a wine can taste hotter after dealc than before. That is the sweetspot phenomenon. Years ago, Fresno State did a brilliant study on this. A syrah fermented dry @ 18% (yep). Dealc’d and blended every .1% down to somewhere in the 13s. Big panel strongly agreed on 4 sweetspots. Many of the in betweens tasted “hot” or other unpleasant descriptors.

    I have consulted on literally hundreds of sweetspots for my clients and an interesting phenomenon is that, regardless of variety, there is a “hot” taste right below a sweetspot and a “thin and saline (usually white)” or sometimes “thin and astringent” just above a sweetspot.

    This is obviously somewhat subjective, but the level of agreement in situ is overwhelming.

    Incidentally, I don’t think your idea of dealcing your topping wine will be best. Vacuum distillation leaves the wine at about 4.5%. You can hit the wine with egregious amounts of SO2 to keep it stable for awhile, but this obviously presents some challenges and limitations.

    Incidentally, you may still be right about some of the higher alcohols coming through in higher quantities with some methods. The method often just called “RO” is not just RO. The RO permeate is distilled and rejoined with the wine (this is the only legal method of using an RO to reduce alcohol). The higher alcohols pass the membrane of the RO into the permeate at about the same rate as ethanol.

    However, in any distillation column there is a subjective cut made between the “congeners” that come off first and last and the “spirits” or ethanol that comes off in the middle. In other words, it may depend on who is operating the still that day. Vacuum distillation and spinning cone distillation obviously are distillations also, whether it is your wine, or permeate, that is being distilled.

    [Shameless plug admission] Our process for alcohol removal is entirely membrane based. The membrane extracts alcohols in virtually the same proportions as they are present in the wine. And it extracts virtually nothing else. It does this without heating the wine at all, and without applying pressure (a few psi to move the wine). So there is no organoleptic impact other than alcohol removal.

    We also offer a sweet spotting service for lab scale trials before you do the work. So you can decide what the best alcohol level for that wine is, or even if you want to do the work at all.

    We also have the ability to work small volumes and limited minimum charges to make it viable for small wineries like yourselves.

    Thanks again for your thoughtful coverage of a complex and important aspect of winemaking.

    Bob Kreisher
    Mavrik North America

    1. John M. Kelly Post author

      Hi Bob – appreciate your weighing in here, shameless plug and all ;-). I did a sweetspotting trial years ago with one of your competitors and am familiar with the effect, and when I was at Vinquiry I had the opportunity to evaluate many de-alc’d wines.

      What I have noticed is that the more stressed the ferment and the higher the alcohol of the starting wine, the hotter the de-alc’d wine seems regardless of the final alcohol, whether it is in a sweet spot or not. Admittedly my experience is limited, and there is more than a small chance that I have an expectation bias.

      RE: my in-house experiment, the plan is to vacuum distill one keg at a time and top with the de-alc immediately. We are not making any effort to capture the “distillate” which will vent to atmosphere through the vacuum pump. Also we will never change the tax class from the level the wine was produced at. What I’m hoping to minimize is incidences of bottling a wine at 14.9% that was produced at 13.5%, which is an all-too-common occurrence for us.

  4. Pingback: Terroirist: A Daily Wine Blog » Daily Wine News: Ribolla Gialla

  5. Mark

    Hey John-Fascinating site, really I mean that. I found you through a comment you left over at Vinography. Anyway, as someone still truly only learning about how wine actually gets made, I thought the part about the yeasts was interesting. We talk to plenty of winemakers and it seems that only a select few even consider letting the natural yeasts take their course. I thought that was both interesting and insightful that you allow them a chance at first.

  6. Tyler Thomas


    Nice summary, good to get you thoughts on this. My experience corroborates much of yours though I have only limited exposure to d-alc’ed wines through friends. Have a wonderful harvest, we’ll be rolling here quite soon!

  7. Pingback: Practical Advice On Managing Wine Alcohol Levels | New World Winemaker Blog

  8. Bob N

    Fascinating, in-depth discussion.

    I’m mostly interested in preventing the increase of alcohol that results from topping w/ wine. It seems to be *unspeakable* to consider replacing the lost water with — omigosh! — PLAIN (distilled) water, rather than wine. I see some potential issues with pH, but can someone explain what the other ‘cons’ would be? At first blush, it seems to make a lot of sense to me. (I’m talking about a situation where you already have a deep, well-extracted wine with pretty good acidity but significant alcohols, say 15% and higher.)


    1. John M. Kelly Post author

      Bob- What may be unspeakable in an ideological discussion may be quite common in pragmatic operations. Barrels in dry cellars are often topped with water to prevent the alcohol level from increasing over time, as the water moves out of the barrel faster than the alcohol. There is no problem with pH as the concentrations of the buffering components are not changing.


Leave a Reply

Your email address will not be published. Required fields are marked *