While it may be said that food safety and food quality are not necessarily interrelated, get lax about either, and you probably won’t have a brand for long. Various sensors and analytical tools can be used to continuously monitor parameters that affect food safety and quality. These sensors and analytical instrumentation can range from thermocouples for temperature monitoring to chromatographs and spectrophotometric devices for precise chemical measurements.
Inline monitoring refers to applications where sensors are mounted directly in the process flow. Inline sensors can be flow, pH or viscosity devices mounted in a pipe; color sensors monitoring French fries for defects exiting a fryer; or an X-ray or metal detector checking pouches at the end of a filling line.
Other types of monitoring, online and at-line, are less direct, but may use the same or more sophisticated instrumentation. For example, an online monitoring system may divert some of the process flow in special piping where sensors are located, while at-line monitoring often refers to a location in the plant where samples are manually taken and either checked at the same location with dedicated test equipment, or taken to the lab for further testing.
While FDA does not yet require electronic recordkeeping, today’s multi-parameter (also known as multivariate) digital sensors and analytical instrumentation feature digital outputs, so why not use them? Besides proving a CIP system has met critical wash/rinse temperatures, flow rates and pH—or that a critical control point’s temperature in a processor’s HACCP plan was met—this data can also be directly used to control the CIP system or process in real time. It can also be applied to a statistical process control (SPC) system to warn when the process goes out of bounds and reveal why it went out of kilter. SPC systems can show what inputs create an out-of-control system, but most importantly, they can help improve food safety, efficiency and quality, and they can reduce scrap and rework.
Roles for inline sensors
Temperature has probably been the most measured process variable, and it’s no different in the food and beverage industry. Omega Engineering, often recognized for its Temperature Handbook, sees a number of its temperature and pressure sensors go into CIP applications, according to Ron Desmarais, senior thermal technology engineer. Measuring temperature in a wide variety of food and beverage processing applications as well as the storage of bulk fluids and materials is also important. But temperature monitoring is often combined with flow and pH in milk processing and pasteurization, and load cells are often added in cooking kettle applications to verify ingredient weights.
Another pH application is inline pH control in carrageenan manufacturing, according to Dave Anderson, Rosemount Analytical marketing director. In another inline application, ozone sensors are used routinely in bottled water filling plants to monitor appropriate germicidal action levels in CIP processes and to monitor any remaining ozone levels in the product. In CIP systems, inline conductivity sensors monitor the efficacy of cleaning solutions.
Other common sensors for inline applications include vortex, Coriolis and magnetic flow meters; level; ORP; resistivity; and dissolved oxygen, according to Joe Downey, Invensys analytical product manager, measurement & instrumentation. Conductivity sensors can be used to control brine density for vegetable quality grading, caustic concentration in pretzel blanching, salt concentration in canning and caustic concentration in vegetable peeling applications. Conductivity sensors can also be used to detect interfaces in juice production. Digital Coriolis flow meters, which also monitor temperature and mass flow, can handle flow measurement for batching from an empty line, monitor CIP flows and track custody transfers during tank truck loading and unloading.
Ola Wesstrom, Endress+Hauser USA senior industry manager, food and beverage, lists some other applications for various sensor types:
- Density, Brix (beverages, HFCS, etc.), Plato (wort quality in breweries), Baumé, salinity (brine concentration for cheese, meat curing, flotation sorting baths) and rate of fermentation
- Percent concentration (solids, alcohol, etc.)
- Non-glass pH sensors for monitoring ascorbic acid addition to acidified foods
- Dissolved oxygen sensors for fermentation and final check prior to packaging of oxygen-sensitive products (beer, wine, other beverages)
- Chlorine sensors for water treatment applications and vegetable and egg processing cleaning installations
- Turbidity sensors for juice and beverages and phase separation
- Color sensors to monitor juice, beer and soft drinks for color consistency and to ensure the correct product is packaged.
UV-VIS-NIR fiber optic-based inline sensors/analyzers can provide real-time data for process monitoring and control. Primary application areas involve color, concentration and suspended solids (turbidity), according to Matthew Rice, Kemtrak AB general manager. Concentration can be used to monitor alcohol, water, sugar, disinfectant strength, blending and dosing control and leak detection of active product or coolant. Color sensing can be used for quality monitoring and controlling of oil color and beer/wort/whiskey color; dosing colorants and flavorings; and checking product consistency. NIR can be used in wastewater monitoring and applications with very high suspended solids concentrations, e.g., dairy (interface detection, product differentiation and waste reduction) or brewery applications such as yeast concentration and centrifuge output solids concentration.
The use of light to identify change or consistency is becoming more commonplace due to its precision, repeatability and ease of use, according to Al Worley, optek-Danulat brewing and beverage team leader. optek has focused on the development and support of high-performance inline photometric analyzers for liquid and gas-processing applications. These include turbidity meters, colorimeters, UV absorption and pH and conductivity sensors that are used to monitor concentrations of milk fat, solids, starch, yeast, color, sanitizers and various constituents.
Benefits of inline monitoring
Today’s automated CIP systems take advantage of several inline sensors, e.g., temperature, flow and conductivity. CIP systems make food safe through the effective washing and rinsing of equipment, but there is no point to overdosing expensive chemicals. For example, the amount of disinfectant used on a hydro cooker for canned food needs to be closely controlled to ensure food safety, since overdosing can cause corrosion and a waste of chemicals, while too little can compromise food safety.
Wesstrom recounts the story of one processor that was able to get its use of disinfectant under control and save some money as well. This particular facility monitored disinfectant by taking grab samples to a lab for analysis twice an hour, a practice that wasted time and proved not to be all that accurate.
Inline analyzers were installed to measure free chlorine, pH and conductivity of the disinfectant, adds Wesstrom. “Real-time measurement saved $13,000 annually in disinfectant costs by eliminating overdosing. These measurements also allowed the automation system to add makeup water based on measured values, saving on heat energy and water usage, and producing less wastewater.” In addition, the inline analyzers eliminated the need to send a lab worker to the cooker twice an hour to take grab samples. The bottom line, according Wesstrom, was a payback period of seven months.
Measuring moisture, content and fat levels—all quality parameters—in powdered milk has often been a “check after you make it” as opposed to having real-time control over the process in the first place. Knowing final protein levels is essential, for example, when processors ship to protein-deprived areas of the world or price powdered milk based on protein levels. In the drying process, the type of spray nozzles must be selected to suit the milk being treated. To get the final product on spec requires determining all key product-relevant parameters simultaneously within the process, rather than conventional manual measurement methods.
One application, implemented by Sartorius, uses NIR spectroscopy to measure all these parameters inline and in real time. Using the Model PMD500, product parameters such as moisture, protein, fat and lactose content are determined quickly and simultaneously without touching or damaging the product. The device uses a combination of sensor types—from color spectrometers and video cameras to an NIR spectrometer—all enclosed in a single housing. In a European application, optimizing the moisture content by 0.1 percent for a product worth approximately €400 per metric ton meant an increase in profit of €400,000 at a production rate of one million metric tons per year.
While Coriolis flow meters were first employed in high-value products (e.g., chemicals, petroleum, edible oils, etc.) for custody transfers, their capability of measuring several process variables at once can be a real benefit to the food and beverage industry. Wesstrom describes an example where this high-end device improved product quality and saved ingredients. In an online application (with a continuous bypass line) of batter in a mixing tank, a Coriolis flow meter was set up to measure viscosity. The batter, consisting of flour, water and additives, is mixed until the correct viscosity is reached, and then pumped to the production tank for processing. The resulting savings in ingredients and the improvement in product quality paid for the installation in less than six months.
Paths to food quality
While the food and beverage industry has worked hard to define HACCP plans and get food safety down pat, it has been lax in quality—largely due to lack of instrumentation in the right places. “The food and beverage industry understands the role quality plays in the value of the brand, and that preserving the highest levels of brand integrity is key,” says Niels Andersen, Invensys vice president, manufacturing consulting. “But the food and beverage industry suffers from a lack of instrumentation in general, which creates a lack of visibility into what is happening in the process and the resulting inability to improve anything that cannot be measured.”
Andersen describes quality improvements at a brewery that sound similar to FDA’s process analytical technology (PAT). “A large international brewer took early measurements in the fermentation process and used these measurements, in collaboration with an Invensys simulation model, to determine if there was scope to make adjustments early in the batch process to improve the end of batch product KPIs.”
The original intent of FDA’s PAT was to set up methods/procedures for monitoring and controlling in real time all process variables measured through inline sensors associated with a new batch. This requires finding the correct process parameters to bring about a batch that serves as the model for succeeding batches. The advantage early on to drug companies was that the process almost always guaranteed perfect results as long as the process was kept under control. Once the perfect, model batch was developed and defined, it then served as the necessary documentation FDA kept as records for the manufacture of a specific drug. This eliminated excess paperwork and recordkeeping for both FDA and pharma companies, and provided consistency in the product. The downside for pharma companies is that if they want to be innovative and make changes to a process to update it or change ingredients, they have to go through the entire documentation procedure again.
“In general, the food and beverage business has been a lot more innovative than the pharmaceutical industry when it comes to using inline measurements to determine the quality of the final product,” adds Andersen. “The reason is that unlike the pharmaceutical industry, the food industry has not been held back by regulations that interfere—because quality deviations have not had a life-threatening impact.”
“Traditionally, inline sensors have had the advantage of being able to do 100 percent inspection of a company’s product,” says Frank Tappen, DataNet Quality System vice president of solution delivery. “They are very good at sorting the bad products from the good. They’re also great at tallying up the number of bad products produced in a particular run, shift, day, week, etc., and providing those statistics.” Tappen suggests that a complete quality system should include SPC software to capture, store and analyze the inline testing results for long-term improvements. “Any inline sensor that records a numeric value for pH, temperature, pressure, etc. could be used by SPC software to monitor for trends, process shifts and other short-term and long-term conditions.”
It’s all intelligence
Generally, PAT and SPC are tools using knowledge derived from measured process parameters to get a process under control. Steve Wise, vice president statistical methods, InfinityQS International, likes to use a more inclusive term, manufacturing intelligence (MI), which provides a broader view of the manufacturing process. Jeffrey Cawley, vice president of industry leadership at Northwest Analytics, also likes the term, MI. “An important implication of MI is that it enables control system results to be integrated with the rest of the manufacturing management systems,” says Cawley. “It gives process engineering a clearer voice in plant and corporate manufacturing management.”
While Cawley has not seen any processors interested in a full-scale PAT implementation, he does report that processors are beginning to experiment with multivariate SPC for process analytics and improvement—most still in the pilot stages. “We work with data captured from a variety of sources,” says Wise. “But the data typically falls into one of three categories: process inputs, process settings and process outputs.” For example, Wise has been working with a French fry manufacturer to come up with a consistent product. To do this, blancher temperatures, time in the blancher, starch extraction, sugar input and many other settings are adjusted for each raw potato lot. As the fries come off the end of the line, they are tested for defects, length ratios, color, texture, flavor, texture and percent solids. If the process outputs are not desirable, changing the line settings may rectify the issues.
“An important implication of MI is that it enables control system results to be integrated with the rest of the manufacturing management systems,” says Cawley. “It gives process engineering a clearer voice in plant and corporate manufacturing management.”
Wise says his company’s MI solution focuses on the concept of a quality hub, a centralized data store or database, to house all quality-related data throughout the manufacturing process. “In addition to capturing inline process data, our food and beverage clients also need that data to play well with their pre-op, SSOP, HACCP and quality data. All these streams of data work together to provide a holistic view of the manufacturing environment.”
Snak King is one processor that has implemented InfinityQS International’s ProFicient MI software. “Before we had an SPC solution, our process quality would have quite a bit of variability,” says Mark Shields, Snak King vice president of quality and innovation. “Our process operators were not really using true SPC decision-making rules, and we found that we would have swings in our product quality. This had a negative impact on quality consistency and increased the amount of product waste we had. We ended up having product quality holds and process aborts.”
Snak King evaluated three different SPC solutions and settled on InfinityQS. Once the ProFicient MI software was installed, the company had a 30 percent reduction in complaints, but there was more. “In areas where we have great data on our variability, we can measure variability in our processes, so we can re-engineer their components and further improve our product quality,” adds Shields. “After implementing Infinity, we were able to reduce our product waste by over a million dollars compared to the previous year.”
Take action, use the tools
There’s an old axiom in the controls industry, “If you don’t measure it, you can’t control it,” and it still applies today. If you have no idea what’s going on in your process because you don’t have sensors in the right places, you can’t expect to compete on food safety, quality and pricing. “A lot of decisions are being made doing the same thing that was done ‘the last time,’ without knowing exactly why,” says Andersen.
“Instrumentation is a lot easier to implement than automation, and when the processes are better understood, you have a great gateway into automation,” adds Andersen.
Bottom line: Don’t automate just because your competitors are doing it. Automate because you can prove the right things were done to maintain food safety, improve quality and optimize your processes for the best return on the assets you already have.
Inline instrumentation key to lawsuit
Recently Anheuser-Busch InBev (A-B) was accused in class-action suits (recorded in Pennsylvania, California and New Jersey) that it wasn’t delivering the alcoholic content it promised on its labels. While there is no Federal rule stipulating that alcohol content be placed on the label, and state laws vary, inline instrumentation can prove what’s in the beer is what’s on the label. According to the Pennsylvania lawsuit, “Sometime prior to 2008, A-B began using inline alcohol measuring instrumentation, known as Anton Paar meters—technology which allows A-B to measure the alcohol content of malt beverages to within hundredths of one percent (i.e., +/-0.01%).”
The lawsuit goes on to suggest A-B doesn’t use inline instrumentation during the final stage where the higher percentage (high-gravity) brew is mixed with water and CO2 to arrive at a “total alcohol content to well below the percentage stated on its labels.” Mixing high-gravity (e.g., 7.5 percent alcohol content), freshly brewed product with water to achieve an endpoint alcohol content is a common practice among breweries and not unique to A-B. In most cases, beer is monitored inline at this final stage to assure the final amount meets the label requirements. Then offline samples are checked again after being bottled.
Offline tests conducted by an independent lab (White Labs in San Diego, CA) and sponsored by National Public Radio found A-B beers to be consistent with labeling and the Anton Paar instrumentation specifications of +/-0.01 percent. Tests showed that for one of the many samples tested, the alcohol content in this particular sample measured 4.99 percent by volume, within the specifications of the test equipment. Based on the independent testing, the report suggests the suit is unfounded.
According to Josh Boxer, an associate attorney with the Mills Law Firm in San Rafael, CA, the lead law firm in the class action suit, the plaintiffs did not independently test the alcohol content in the beer to confirm whether it was indeed watered down.
For more information:
Ron Desmarais, Omega Engineering, 856-467-4200, email@example.com
Dave Anderson, Rosemount Analytical, 949-757-8500, firstname.lastname@example.org
Joe Downey, Invensys, 508-549-4690, email@example.com
Ola Wesstrom, Endress+Hauser, 888-363-7377, firstname.lastname@example.org
Matthew Rice, Kemtrak AB, +46 10 511 0700, email@example.com
Al Worley, optek-Danulat, 800-371-4288, firstname.lastname@example.org
Steve Wise, InfinityQS International Inc., 703-961-0200, email@example.com
Frank Tappen, DataNet Quality System, 248-357-2200, firstname.lastname@example.org
Jeffrey L. Cawley, Northwest Analytics, 503-224-7727, email@example.com
Niels Andersen, Invensys, 508-543-8750, firstname.lastname@example.org
“Budweiser May Seem Watery, But It Tests At Full Strength, Lab Says,” Dan Bobkoff and Bill Chappell, Morning Edition, National Public Radio; 27 Feb. 2013.
“Chemometrics in Food and Beverage,” Infometrix (www.infometrix.com), 1996.