In this 15 minute webinar, food scientists Mary Galloway and Dr. Zachary Cartwright talk about the theory and application of one-minute water activity readings, including: - the physics-based model and hybrid analytics breakthroughs that make a one minute reading possible - why these techniques work for samples that usually have long read times - comparison to traditional first-order “fast” or “quick mode” readings - pros and cons from Mary’s testing experience in the food lab The webinar will be followed by a ten minute Q & A. Presenters Zachary Cartwright is a lead food scientist at METER Group. He helps customers achieve complete moisture analysis of their products and is an expert in the use of the Vapor Sorption Analyzer (VSA). He has a Ph.D. in food science from Washington State University and a bachelor’s degree in biochemistry from New Mexico State University. Mary Galloway is the application scientist for METER Food and manages the Food Research & Development lab. She is a contributing author on several publications concerning water activity and its influence on physical properties. Mary uses her years of experience to help customers understand and solve their moisture-related product issues.

2. HOW TO MEASURE
WATER ACTIVITY
IN ONE MINUTE
MARY GALLOWAY ZACHARY CARTWRIGHT, PHD
APPLICATION SCIENTIST LEAD FOOD SCIENTIST
METER GROUP, INC. USA

3. Zachary Cartwright, PhD
Lead Food Scientist
Mary Galloway
Application Scientist

5. CONTENTS
THE SCIENCE BEHIND WATER ACTIVITY MEASUREMENTS
KEYS TO SPEED
USING SPEED TO REDUCE UNCERTAINTY
IMPACT OF FAST WATER ACTIVITY
KEY TAKEAWAYS
Q&A

6. THE SCIENCE BEHIND
WATER ACTIVITY
MEASUREMENTS

7. Several first-order processes,
each with a different time
constant
SAMPLE
CHAMBER
SENSOR
Not only are time constants different
from each other, they’re different for
different products and can change
under different conditions
FROM A PHYSICS PERSPECTIVE

8. KEYS TO SPEED

9. KEYS TO SPEED

10. FIRST-ORDER MODEL
Well-known engineering strategy
Speed up reading by using current data to extrapolate

11. FIRST-ORDER MODELING
Using reading time series data
• Obscures several different processes
• Need to wait for fast processes to be
nearly complete to fit data.
Takes much longer than
one minute
• Results from this method are
unsatisfactory (errors unpredictable)

12. FIRST-ORDER MODELING
METER Competitor A Competitor B
Aw Time (min) Aw Time (min) Aw Time (min)
Equilibration Fast Equil Fast Equil Fast Equil Fast Equil
Potato Chips 0.099 7.2 0.104 0.078 10.5 15.2 0.094 0.106 4.3 15.4
Peanut Butter 0.194 5.8 0.134 0.120 31.1 111.4 0.212 0.222 4.0 16.4
Ground Black
Pepper
0.365 4.7 0.423 0.370 32.5 11.6 0.428 0.368 3.4 19.9
Dog Kibble 0.425 3.2 0.414 0.419 14.5 26.6 0.408 0.414 3.8 14.9
Craisins 0.504 4.3 0.502 0.500 9.8 18.4 0.522 0.504 3.4 7.9
Beef Jerky 0.716 3.2 0.702 0.696 13.2 16.2 0.712 0.700 3.7 26.4

13. FIRST-ORDER MODELING
Chilled Mirror Competitor A Competitor B
Aw Time (min) Aw Time (min) Aw Time (min)
Equilibration Fast Equil Fast Equil Fast Equil Fast Equil
Potato Chips 0.099 7.2 0.104 0.078 10.5 15.2 0.094 0.106 4.3 15.4
Peanut Butter 0.194 5.8 0.134 0.120 31.1 111.4 0.212 0.222 4.0 16.4
Ground Black
Pepper
0.365 4.7 0.423 0.370 32.5 11.6 0.428 0.368 3.4 19.9
Dog Kibble 0.425 3.2 0.414 0.419 14.5 26.6 0.408 0.414 3.8 14.9
Craisins 0.504 4.3 0.502 0.500 9.8 18.4 0.522 0.504 3.4 7.9
Beef Jerky 0.716 3.2 0.702 0.696 13.2 16.2 0.712 0.700 3.7 26.4

14. SECOND-ORDER MODEL
Uses complex modeling process
• Second order model developed by physicists
• Uses strategies from predictive environmental modeling
Requires sample-specific coefficients
• Sample needs to be identified so correct coefficients are used
Needs 3 readings of a specific sample

15. SECOND-ORDER MODEL
0
0.005
0.01
0.015
0.02
0.0000
0.0928
0.1130
0.1500
0.1790
0.1793
0.2500
0.3877
0.5000
0.5126
0.5385
0.7406
0.7600
0.7775
0.8291
0.9200
0.9339
0.9840
1.0000
WaterActivityMaxError(aw)
Expected Water Activity (aw)
Fast aw vs. Final Dew Point aw
One-Minute Aw Max Error Final Dew Point aw Max Error

16. USING SPEED TO REDUCE
UNCERTAINTY

17. ONE-MINUTE
READINGS POSSIBLE…
“Normal readings” already fast
• Readings take typically a couple of minutes
Clean instruments
• Eliminate materials that will adsorb water
Accurate instruments
• second order modeling can be applied to make faster
If your sensor is slow, you can never get a
one-minute reading.

18. VARIABILITY > ACCURACY
VARIABILITY IS A
BIGGER ISSUE
THAN ACCURACY
More measurements will
reduce uncertainty better
than increased sensor
accuracy

19. Inherent variability makes
uncertainty HIGH
10 MINUTES
1 READING
UNCERTAINTY =
𝑺𝒕𝒂𝒏𝒅𝒂𝒓𝒅 𝑫𝒆𝒗𝒊𝒂𝒕𝒊𝒐𝒏
𝑵𝒖𝒎𝒃𝒆𝒓 𝒐𝒇 𝑹𝒆𝒂𝒅𝒊𝒏𝒈𝒔
UNCERTAINTY
1
0.3
HOW UNCERTAINTY IS REDUCED
FOOD
SAMPLE
10 MINUTES
10 READING

20. HOW UNCERTAINTY IS REDUCED
Fast water activity allows you to include more readings in your average to
drastically improve uncertainty.
If you can get 4 readings, that will cut uncertainty in half.
The main source of uncertainty in factories is not taking enough readings.
The inherent variability in the product means that one to two readings… is
nowhere near enough to characterize aw.

21. IMPACT OF
FAST WATER ACTIVITY

22. IMPACT OF
ONE-MINUTE READINGS

23. BUSINESS IMPACT
A well-known cured meat
manufacturer:
-Clearing their ovens and cure
rooms in a fraction of the time.
-40% increase in production
throughput (no additional
investment).
-Saving over $1M in planned
investment to scale operations.

24. BUSINESS IMPACT
A prominent oven baked oat bar provider
-6 readings went from over an hour to <10 min.
-Know with certainty when they can move to packaging.
-No unnecessary rework or lost batches.
An innovative sunflower seed butter producer
-30 min test down to 1 min.
-No longer having delays on batch releases.

25. KEY TAKEAWAYS

27. QUESTIONS?

28. METER GROUP USA, INC.
2365 NE Hopkins Court
Pullman, Washington 99163
Phone: (509) 332-2756 / (800) 755-2751
Fax: (509) 332-5158
Email: mary.galloway@metergroup.com
zachary.cartwright@metergroup.com