Development of tools and procedures to improve the consistency of farmstead cheeses

Objective 1:  Using model cheeses determine the key measurable and qualitative parameters for defining the optimal characteristics of the cheeses at the time of consumption. Three different types of cheese will serve as model cheeses: a raw milk, washed-rind semi-soft cheese, a raw milk gouda, and mozzarella curd that is frozen for future stretching.

As planned in project proposal: An initial sensory evaluation session will be held with each cheesemaker to discuss the optimal flavor and texture characteristics of the cheeses. The cheesemaker will provide several cheeses that represent a range in quality to determine control limits. The cheeses will be evaluated by the cheesemaker, the Penn State researchers, and others that may assist the cheesemaker with routine evaluations.

A lexicon of flavor and texture descriptors will be developed for each cheese, using methods described by Drake et al. (Development of a Descriptive Language for Cheddar Cheese. 2001. J. Food Sci. 66(9):1422-1427). Descriptors will be drawn from those found in the literature and used in cheese competitions for similar cheeses. Score sheets will be developed for the cheesemakers to quantitatively assess the sensory aspects of the cheeses during the study.

For the washed-rind, semi-soft cheese and the gouda cheese, determination of other measurable quality indicators will be based on a discussion with the cheesemaker on the cheesemaking process and current problems. Variables to monitor include the pH and moisture content of the cheese at different stages of aging, temperature and humidity of the aging facility, and sensory attributes. Cheeses chosen for the initial on-site evaluation will be measured for these markers to determine the control limits for optimal quality.

The key parameter for stretching mozzarella is a curd pH of 5.1 to 5.3. The mozzarella curd is sold frozen, to be thawed and stretched by the end user. At the initial session, the manufacturing process will be evaluated to determine when pH measurements should be taken and to see if additional monitoring of the process is needed.

Project activity: A master list of sensory descriptors was developed that encompassed appearance, flavor and aroma, and body and texture attributes associated with all types of cheeses. We developed sensory ballot templates with various types of attributes scales, such as just about right (JAR), intensity, and difference from control. These were used to select attributes and develop sensory ballot scales that were appropriate for each cheese. We developed a sensory evaluation guide to assist the cheesemakers with conducting sensory evaluation on their cheeses in their own facilities.

We had an initial meeting with the first cheesemaker to evaluate cheese and define measurable quality parameters – sensory attributes, make procedure targets (pH, time, temperature), and aging room conditions (humidity, temperature, time). We discussed and reviewed the initial sensory guide, ballots, and data collection sheets, which were then refined to better meet their needs and reflect actual practices. They agreed to send us cheese samples and production data approximately every two weeks. This did not work out as expected due to their schedule, and we received samples approximately every 3 to 6 weeks. The data still provided a picture over time, just not with as many data points as we had anticipated.

We met with the second cheesemaker to evaluate cheese and define measureable quality parameters. They changed some of their procedures and were not as interested in as much assistance as originally planned for the project. We reviewed the sensory guide, and production data collection sheets, and modified the sheets as needed for their use. Their primary interest was following the cheese process and the final pH of the cheese, so we tracked their milk composition and processing data. They were not interested in cheese composition analysis.

We had anticipated that environmental sampling and sanitation tracking would have been an important factor in cheese quality, and based on these initial visits that did not appear to be the case for these two cheesemakers.

We added a third cheesemaker to the project in December 2015. This cheesemaker had been working with Penn State to resolve some quality issues on a Gouda style cheese. Bringing them into the project seemed to be a good fit since we were not using as many analytical resources as we had anticipated with the other cheesemakers, and adding another cheese was not that much effort with regard to data collection templates and tracking spreadsheets. An on-site visit was made in mid-December to refine a sensory ballot and cheese processing and aging data collection sheets. Key measurable parameters for this cheesemaker were focused on several sensory attributes, and collecting more complete production data to provide a basis for tracking quality and troubleshooting. They began sending cheese and data to us in January 2016 for evaluation and tracking. The initial visit showed there was some sanitation issues that might have be an important influence on the cheese quality, and these were discussed in following on-site visits.

The input from all cheesemakers on how they evaluated their cheese and identified the attributes and evaluation scales that best suited their needs was extremely valuable in guiding how we developed the data collection worksheets, tracking spreadsheets, and the sensory evaluation guidance documents for our cheesemaker collaborators and for the templates we developed for all cheesemakers.

Objective 2 - Develop a tracking system to record the milk composition and quality measurements, processing and environmental conditions, and cheese characteristics.

As planned in project proposal:The batch sheets currently used by the cheesemakers to record the cheese making process and other factors will be reviewed. Modifications to these forms or development of new ones will be made as needed to ensure that all key variables are recorded. Worksheet templates will be developed using Microsoft Word.

Microsoft Excel spreadsheets will be developed to record milk composition (fat, protein, lactose, solids content) and quality parameters (standard plate and somatic cell counts), cheese making process variables (pH, times, temperatures), environmental conditions (temperature, humidity), and cheese quality characteristics (flavor and texture score, pH, moisture content). Graph templates will be developed within Excel to monitor and compare trends in raw material and process variables with finished product measurements.

Standard procedures will be developed for data collection by the cheesemakers and for secure transfer of data to Penn State for analysis. The procedures will be shared with the cheesemakers during an on-site visit to ensure that they understand the procedures and how and when to collect and report the data.

Project activity:  During our initial meetings with the cheesemakers, we discovered that the data they were capturing during the cheese making process was lacking in what we needed to get a more accurate picture of their process. So we also developed cheese making data sheets to capture key information such as time, temperature, and pH targets.

Multiple data collection worksheets (Word) and spreadsheets (Excel) were developed and customized to capture and track data on milk composition, cheese composition, sensory evaluation, and processing and aging room parameters for each cheesemaker. The documents were provided to the cheesemakers as soon as they were developed after the initial meetings. Refinements were done as needed based on the cheesemaker’s feedback as they used the documents.

These documents were customized to the cheese characteristics, processing steps, and other specific needs for the 3 cheesemaker collaborators. There were many common features of the documents that provided the foundation for the generic templates developed in Objective 5 that are applicable for use by any cheesemaker.

Objective 3 - Track milk composition and quality data, processing and environmental conditions, and cheese characteristics over one year to capture seasonal effects.

As planned in project proposal: The cheesemaking process, environmental, and cheese quality data will be collected by the cheesemakers and sent to Penn State according to procedures and worksheets developed in Objective 2. Milk and cheese samples will be shipped periodically to Penn State for analysis of the parameters determined in Objective 2. The cheesemakers receive data on milk composition from their dairy cooperatives and will share the information with Penn State for tracking. All data will be tracked in Excel spreadsheets by Penn State for one year to capture seasonal trends.

One month after data collection begins, the researchers will visit the cheesemakers to discuss and observe the data collection process, and refine worksheets and procedures if necessary. The researchers will collect milk and cheese samples for analysis at Penn State. Four additional visits will be made to the cheesemakers over the course of the study period to observe and collect data and samples under different seasonal conditions. At each on-site visit, Penn State will sample the cheesemaking and aging room environments using ATP swabs as an indicator of cleaning and sanitizing practices. 

Project activity: The intent of the project was to collect data on raw milk, cheese making, aging, and environmental variables regularly throughout the year to track variation. We expected that the cheesemakers would collect cheesemaking data on a daily or weekly basis, and send cheese samples to us approximately every 2 weeks for compositional analysis. Additionally, there were four visits to cheesemakers included in the project to observe data collection, collect samples, and swab facilities for sanitation indicators.

We did not collect as much data as we proposed because of the realities of working with small-scale cheesemakers. We learned in the project that our initial goals involved an unreasonable amount of things for cheesemakers to track on a daily basis because of the multiple demands on their time and the need to focus on their commercial commitments.

We learned at the beginning of this data collection period that there were more issues with the cheesemakers consistently tracking their process and collecting sufficient data than any impact of the environmental conditions. Therefore, we decided not to conduct the environmental analysis as we had planned.

All data received was entered into the data tracking spreadsheets that were customized for each cheesemaker: milk tracking, cheese make data, post-make day processing, cheese composition, sensory analysis, and cheese tracking summary. Cheese composition analyses were done at Penn State and included pH, moisture content, salt content, salt-in-moisture, and fat content.

By June 2016 we had limited data to analyze. One cheesemaker had less than 15 instances of complete data, another had only made several batches of cheese, and another sent us a lot of cheesemaking data, but not much of the sensory data. Therefore, we had limited data on the key quality measurements to compare with the cheesemaking variables. We contacted the cheesemakers to see if they could send us more complete data to help us with the analysis. We received a little more data in the final 2 months of data collection, which gave us some endpoint data that we needed for the comparisons.

By the end of the data collection period in August 2016, we had full cheese compositional data analyzed by Penn State on 28 cheeses for one cheesemaker and on 21 cheeses for another cheesemaker, but there was sporadic make day data, aging data, and sensory data for these two cheesemakers. For the third cheesemaker we had data on milk composition data on 191 samples, 101 days of cheese make data, sensory data on 28 cheeses, and no compositional data on the cheeses. This pattern of data collection impacted our ability to complete Objective 4 as planned, but gave us valuable information on the real world expectations for small-scale cheesemakers.

Objective 4 - Correlate manufacturing variables with cheese quality to determine which variables are most important, and make recommendations on how to adjust the cheese making process to compensate for the variations that influence consistency.

As planned in project proposal: All process and environmental variables will be graphed in Excel to determine trends over the course of one year of cheese making and aging. These trends will be qualitatively correlated with cheese quality parameters to identify the variables that are most influential on cheese consistency. Information published in the scientific literature will be reviewed and summarized to provide straightforward recommendations on how to adjust the cheesemaking process to compensate for the key variables identified.

Process and composition data will be collected for an additional 6 months following the implementation of the recommendations. The data will be analyzed as above to determine if changes in practices resulted in improved quality and consistency of the cheese.

Statistical analysis based on replications of individual cheesemaking days will be discussed with the Penn State Statistical Consulting Center, but may not be feasible due to issues inherent in following commercial farmstead cheesemaking operations compared to a controlled laboratory study. The ability to gather replicates based on seasonal variation is beyond the scope of this project.

Project activity: We received some data for the cheesemaking process, some for milk and cheese composition, but did not have much sensory data on the finished products. This was a problem because it was difficult to analyze the effect of manufacturing variables without a measurable endpoint for comparison. Because of the lack of continuous data, we were not able to offer specific instructions to improve their procedures, or provide any suggestions based on seasonal changes in their milk supply. We did the best we could with the data that we had, and were able to identify quality trends in the data and identify areas of inconsistency in their general processes and the finished cheeses for all cheesemakers.

The cheesemaker meetings were delayed from the July to September time frame until an October to January 2017 time frame due to scheduling issues. We delayed the meetings a bit to give us time to work on the generic templates in Objective 5, so that we could provide these to the cheesemakers at the data review meetings. We met with two cheesemakers in 2016 and the third in January 2017 to deliver the data, review the trends, make some minor suggestions and show them how to use the worksheets and spreadsheets.

We intended to track improvements in cheese quality and/or processing for 3 months after providing feedback from the data analysis. But based on the data we received during the collection phase, we were not able to make specific recommendations for changes to their manufacturing process that we could track for the next period. Many of the trends we identified were more general processing changes that can’t be quantified in a short period of time, so we ceased data collection. Because of the lack of consistent, replicated data we did not conduct any statistical analyses on the data.

The cheesemakers appreciated the feedback and began making small adjustments and paying more attention to collecting complete data so they can do their own tracking. The data collection worksheets developed for the make day and aging processes were useful to help them know what data needs to be taken for tracking cheese consistency.

Objective 5 - Generate fact sheets and data tracking templates that can be modified by other cheesemakers to meet their specific needs.

The protocols and worksheets used in this study will be modified to remove information that is specific to the study collaborators. Generalized guidelines for quality determination and process control will be published as fact sheets. Data collection and tracking templates will be made in Word and Excel so that they can easily be modified by any cheesemaker to meet their needs. These materials will accessible to all cheesemakers from the Penn State Dairy Processing website.

This project was conducted to help small-scale cheesemakers make more consistent, high quality cheese and be better able to troubleshoot manufacturing problems. Normal variations in milk composition, the cheesemaking process, and facility conditions affect the quality and consistency of cheese. Setting up a program to collect and track these variables and evaluate their influence on the cheese poses challenges for cheesemakers that have limited time and resources.  Our objective was to develop a system of easy-to-use worksheets and spreadsheets that could be customized to a cheesemaker’s needs and provide an infrastructure to track cheesemaking variables and relate them to cheese quality attributes.

Project activity: To accomplish this goal, we worked with three cheesemaker collaborators who made different types of cheese, and provided us with a range of cheesemaking variables to use in the development and refinement of the cheese tracking system. We developed a customized system for each cheesemaker because the desired attributes of the ideal cheese and the selection of key raw materials and processing parameters to be measured to track their influence were dependent upon the type of cheese being produced and the wishes of the cheesemaker.

The development of the generic templates and instruction guides was started earlier than anticipated in order to give them to the cheesemaker collaborators when we delivered their final data in November 2016 through January 2017. This allowed the cheesemakers to use the documents for their cheeses that were not part of the study and provide us with feedback so that we could refine the documents. During this process we realized that the spreadsheets we developed to track many research variables were too complicated for general use, so we refined the spreadsheets to make them more user-friendly but still provide flexibility for tracking multiple variables.