The beginner's guide to measuring yield and waste in food manufacturing

Continuous improvement and real-time data are the power couple of waste reduction. Data, collected automatically and delivered in real time, powers the Plan-Do-Check-Act cycle by exposing issues, letting you dissect and diagnose, delivering live data to your people so they can monitor and control, and then confirming your results.

Everybody knows you could get that data by investing millions in brand new high tech equipment.  But that’s not the only way to go. Following are examples of manufacturers who are getting impressive results by collecting real-time data automatically from the legacy machines they already own.

See how they’re dealing with seven common offenders in the battle against waste.

Overpack

If you under pack, you’ll have to scrap or repackage. So most manufacturers err on the side of overpacking. But overpack is a direct giveaway. There’s low-hanging fruit here for the food manufacturer that can exploit real-time measurements. For example, a customer running an automated bagger and check weigher noticed that over time, build-up on their equipment would trigger under pack. Adjusting the machine would bring the bags in spec temporarily, but suddenly they would find that conditions had changed and they had been overpacking by 5% for some time.

Data collection: By connecting the check weigher to SKALA control, they can see weights reported in real time on a monitor above the weigh station with control limits clearly marked and out-of-spec readings flagged.

Result: Overpack decreased from 2% to 1%. Savings of $84,000/month.

Over/underprocessing

Processing shelf-stable foods often involves water: baking, drying, dehydrating, rehydrating. There’s a sweet spot nearly every product needs to hit. The ability to get there, consistently, on the first pass, is a powerful way to reduce waste. One customer rehydrates dried fruit to bring it up to an acceptable moisture, then bags it. They can’t measure the moisture until it comes to equilibrium two days later. Final moisture values can fluctuate between 13% and 18% moisture. Failing to meet the 18% spec can mean a 5% difference to the bottom line.

Data collection: This customer requires data from multiple sources. They collect their initial, intermediate, and final (second day post-processing) moisture content measurements automatically through SKALA-connected instruments. Ambient temperature and relative humidity are collected inside the rehydration chambers by using connected temp/RH sensors. Gross product weights are collected before and after processing through a connected balance, and the weights are then used to determine moisture added.

Data collected over time are used to create an algorithm that recommends inputs necessary to bring each batch to 18% moisture based on measurements from an incoming lot of dehydrated product.

Result: Savings of $100,000/month.

Process not right

A baked goods manufacturer scraps significant quantities of product due to inconsistent bakes. By collecting data, they determine that the problem is not in the oven, but in a previous processing step in which a filling was cooked down. Variability in the water activity of the filling is causing nonconforming final product.

Data collection: QC collects a sample every five minutes to measure water activity of the filling at the line. Results are reported on a monitor at cook station and to the line lead. By seeing water activity as it happens, the operator can adjust cook time to keep the filling water activity in control.

Process not consistent

Variability in the production process often leads to waste. One customer finds that variabilities in oven temperatures lead to inconsistent water activities in the final product. By monitoring final water activities in real time, they can adjust processing time to dramatically improve consistency.

Data collection: Oven temperatures in real time, processing time as measured by conveyor belt speed reported in real time on a process control dashboard. Actions: adjusting belt speed to vary processing time.

Result: Process variation reduced by 50%. Savings of $50-80,000/month.

Equipment malfunction

A customer that produces an extruded bar has a guillotine that cuts these bars into sections. When the guillotine starts to malfunction, bar sizes go out of spec and production runs out of control, creating waste. Operators need to be able to monitor the process so they can see trends and make adjustments or call maintenance to fix the guillotine.

Data collection: Net weight and size measured in real time from connected scale and calipers. Data and trends reported on a dashboard co-located with the guillotine and a second dashboard visible to the line operator. When the process goes out of control, a worker can stop the line and adjust the guillotine or get in touch with maintenance immediately.

Result: The customer estimates that they catch problems 50% faster with improved communication channels.

Dropped/lost product

This is the most basic of all forms of waste, and one of the most difficult to track. To do it properly, you have to be able to track yield and waste at many points in the production process. You also have to account for loss that is not waste—water that is baked out of the product, for example. Many manufacturers of shelf-stable products remove water (by baking, drying, dehydrating). They can use water activity to predict expected yield. By combining automated waste tracking and expected yield, it’s possible to determine how much product is dropped or lost.

Variation in raw ingredients

Food ingredients are hugely variable year on year. Growing locations and conditions affect size, sugar content, moisture content, and more. Adjusting your process to account for differences in raw ingredients is the most complex of all challenges. To be successful, you have to understand which attributes of incoming ingredients affect processing and measure them. Real-time data tied to each lot and batch allows you to leverage food science expertise and create a predictive algorithm to recommend adjustments to processing based on attributes of incoming ingredients.

Where to start

There are obviously many places in the production process where you can begin attacking waste. The best implementations start with a single target that has significant potential ROI. After the payback period, part of your savings can fund the next project and power continuous improvement.

New method to stop estimating, start tracking maximum vs. actual production yield

In this 20-minute coffee break webinar, Dr. Brady Carter introduces a new way to calculate maximum production yield batch by batch using a moisture sorption isotherm. It only requires three types of measurements: weight (initial and final), moisture content (initial and final), and water activity (final).

Learn:

• Why water activity provides a unique way to predict maximum yield
• How automated data collection makes these waste calculations feasible
• The results of a study measuring waste in potato flakes produced by a university-run pilot plant
• Challenges and limitations of the method