Vineyard & Winery Management Magazine
Feature Story
Blending Best Practices
Creating a Lab-Scale Blend
Story by Patricia Howe
Blending is a critical aspect of the winemaking craft, and historical texts make much of the general improvement of wine through blending. For example, wines that are flat can be blended with those that are acidic to make a larger volume of an improved wine. Early sensory tests of simple two-component blends always found the blend to be of "higher quality" than the individual components.
While skillful blending can certainly create a wine that's greater than the sum of its parts, the process isn't a simple matter-even for experienced winemakers. As one wine competition judge recently told me, there are a fair number of clunky and awkward blends in the marketplace. That comment intrigued me. What's the difference between a good blend and a bad blend? And how do you make a well-blended wine?
The most likely basis for a "clunky" blend is one of two issues: poor blending and mixing logistics in the cellar, or the poor use of sequential imitative blends produced from different sets of components throughout the year.
This article will review the process of preparing a quality trial lab-scale blend. The first steps are to identify what the finished product is intended to be and define this ideal product using any appropriate method (chemistry, vocabulary, or physical standards). The second is to have the appropriate homogenous components from which to make your ideal wine. Third, you must methodically work through a regimen of lab-scale blends to meet the requisite parameters and then use tastings to choose the final blend. Next, you must have the appropriate group participate in the tastings and agree on what the standard is. And finally, you must compare your new blend with your original standards and verify that all parameters have been met to your satisfaction.
Components and Sampling
A blend is a mixture of individual components, and defining and listing them is your first step. Each component may be from a single vineyard, a single tank, or a single lot in many tanks or barrels. Whatever its volume, each component should be consistent. If a multi-tank component has one tank with a spoilage issue, then you actually have two components. If one vineyard lot went into barrels from several suppliers you may have several lots. At some point logistical limitations may dictate how uniform your components are, but these limitations must be weighed against the possible ramifications.
To produce a lab-scale blend for tastings and analyses, representative samples of each component must be pulled. It is critical that the samples are representative of their components. Barrel sampling is always difficult, and blending multi-barrel samples from each lot is one approach. Most wineries have developed some kind of algorithm for determining how many barrels in a given "lot" must be sampled and combined to provide a wine that is representative of the whole group of barrels. One straightforward method of getting representative components from a barrel lot is to pull aliquots from the square root of the number of barrels in the lot and blend them for a representative lot sample.
For each component sampled, there are several pieces of information you will need in order to proceed. Of primary importance is the volume of each component, which in some cases must be estimated due to the physical state of the wines, as wines on lees or prior to filtration or other processing may change significantly in volume. Depending on the motivations for making a blend-economic, regulatory, chemical or sensory-each component should also have composition attached to it. (By composition I mean the associated cost of the wine, such as grapes, processing, etc.; the percentages of variety, appellation, vintage or other regulated component; the chemical analyses and any added materials; and any relevant comments or classifications on wine attributes, including quality, color, aroma, flavor.)
Calculating Blend Percentages
Many wine tracking programs have blending functions that allow you to make trial blends using the data already in your database. Sometimes it is easier to export your existing data into a spreadsheet and manipulate it there. If your software has an export function, creating a custom blending spreadsheet may be an option. An additional advantage to using a spreadsheet is that a hard copy provides documentation of your ideas and notes. Additionally, a few keystrokes can produce a guide for producing the lab blends and preview mathematically important issues. "Fudge factors" can be applied to account for anticipated processing losses prior to blending.
Most compositional elements will blend in a simple linear manner (with some non-linear exceptions such as pH). Sensory qualities are the most complex. Although the underlying chemicals responsible for aroma, flavor and texture will blend mathematically, their perceived effects do not. Complex chemical interactions involving macromolecules, vapor pressure, van der Waals forces and electronic spin conspire with human sensory systems to create flavors, colors, textures and aromas in the blends that are not necessarily in proportion to those in the components. Early researchers found that sensory estimates of strong aromas followed a geometric or power function. Simply stated, the blends of the best wines are not always the best blends, and blending in a bad wine may actually improve the finished product. No simple formula exists to predict how a blended wine will be perceived, although practice and experience seem to aid -predictions.
Making a Lab-Scale (Trial) Blend
If you have standards (i.e., chemical or accountable parameters, physical samples, or other specifications), have them available for constant reference. Sometimes the standard exists only in the winemaker's head, which is convenient for the winemaker but not for anyone else. Or perhaps you are trying to imitate a particular style and can evaluate flights of competitors' wines to determine a physical reference or descriptive specification.
Next, you need to attain an optimized blend to meet your sensory quality parameters. Realize that you are not only limited by the components that are available but also by the people tasting these blends. Blending and selection of the final blends are part of one of the ultimate mysteries in the winemaking world, and many personalities in this industry will do their best to foster and protect this secretive aspect. Although there is plenty of mystery, science and procedures can get even the most challenged of winemakers close to an acceptable blend. The important resources for this are thoroughness, access to data, and consistent follow-through.
Let us differentiate the art and science of blending from the process of component weeding, which is the oversimplified practice of eliminating wines that individually do not meet a set of parameters. Winemakers who overuse weeding (they may be limited in time, resources or experience) do not make full use of the available components and therefore limit themselves to a smaller subset of wines. These winemakers often do not grasp the understanding of the quality contributions made by adding in wines that may not individually meet a quality standard.
Each winery must consider the relative importance of the quality of the blends to their overall operations and weigh the relative amount of time and effort to spend on this aspect of winemaking. If the decision is made that this is an important aspect, then it behooves the winery to involve several key figures in the discussion of goals, the generation of standards and the provision of blend requirements. Additionally, for consistency and diligence, more than one person should be involved in the sensory evaluation of the final blends and comparison to any existing ideals. In small facilities, this might be a simple winemaking team that blends, tastes and compares the wines to the standards. In larger facilities this might involve not just the production team but also a research or sensory department that will ensure that any differences are minimal or unimportant.
Here are the steps to follow:
1.Evaluate and review your standards specifications and requirements.
2.Evaluate your components. Prepare a list of each component with all related information (volume, chemistry, variety, appellation, barrel issues, etc.). Consider the processing state of each component and determine the true volume available. Taste each wine and record color, clarity, aroma, flavor and quality, along with any notes on appropriate target blends.
3.Blend and evaluate a range of lab-scale trials optimized for all the non-sensory parameters and get a perspective of the resulting range of sensory qualities. Make sensory notes as above for each blend.
4.Begin a systematic inclusion/exclusion of the most variable sensory components and evaluate the effect on the sensory and non-sensory parameters.
5.Tweak the trial blends with inclusion/exclusion of less variable sensory components and check their effects.
6.Take your final trial blends and compare them to your standards. You may want to consider making a larger lab-scale blend and allow it time to "marry." Then, revisit this blend and review your standards after a suitable rest period of several days to a few weeks.
7.Consider any post-blending treatments (such as barrel storage, additions, processing) that might induce differences in the blend and require downstream back-blending.
Obviously, there are many challenging aspects of producing a quality blended wine. In the January/February issue, we will discuss the process of making a lab-scale blend into a large-scale physical reality.
Pat Howe, an enologist and consultant specializing in sensory and chemical analysis, learned how to blend wine as a sparkling winemaker, working with great Champagne blenders from both Moët (in France, California, and Australia) and Mumm (in France and California).