Reducing Fermentation Variation - Welcome to the Master ... · Reducing Fermentation Variation ......
Transcript of Reducing Fermentation Variation - Welcome to the Master ... · Reducing Fermentation Variation ......
District NW Spring meeting 2012
Reducing Fermentation Variation
Peter Bouckaert New Belgium Brewing Co
Factors affecting fermentation variation Variables: • Yeast:
amount, vitality, viability, generation, time of addition,…
• Fermenter conditions: Number of brews, temperature tank, KO temperature, outside temperature,….
• Fermentation heat up rate • Nutrients:
Sugar profile, aeration, minerals, vitamins, FAN,…
Variables yeast’s past: • Yeast storage time and
temperature, • length of previous
fermentation, • fermentation cool down
rate • Fermentation heat up rate • Anomalies
Project Goal
• Problem: Fermentation times are too variable
• Key measure: Standard deviation of fermentation time of beer1 excluding generation 1 fermentations
• Goal: Standard deviation <3 hrs
• Other benefits: Beer consistency, tank turn around time, other beers will benefit from learning on beer1
Project drivers:
• Postponing buying new fermenters – Less fermentation variation, easier planning
• Manufacturing Execution System (MES) implementation – What date to collect from the process?
• Reduce frequency of diacetyl measurement if fermentations are more predictable.
Process:
yeast storage 3C
Propagation Brew
Ferment: Heat up KO --> 20C data for prediction Ferment to end of VDK rest to be predicted Cool down
Fermenter queing
Fermenters available
centrifuge/chill -1C
yeast storage 3C
Propagation
Brew
Ferment: Heat up KO --> 20C data for prediction
Ferment to end of VDK rest to be predicted
Cool down
Fermenter queuing
Fermenters available
centrifuge/ chill -1C
Measurements defined
• Key measure: Beer1 fermentation time from start of first brew coming in the fermenter (Knock-out or KO) until VDK (vicinal diketones) are low enough to cool down.
• Method of measurement: Through process historian of PLC, later transferred to Manufacturing Execution System Except VDK (see next slide)
• Measurement variation: 12 minutes variation between operators - OK
VDK measurement Measurement System Analysis: Operator, range, sample prep
Prediction curve beer1: Brand specific
Source Variance Standard Deviation % Contribution Total Measurement (Gage) 1726.17857 41.54730522 1.15% Repeatability 165.5 12.86468033 0.11% Reproducibility 1560.67857 39.50542458 1.04% Operator 393.904762 19.84703408 0.26% Oper * Part Interaction 1166.77381 34.15807093 0.78% Product (Part-to-Part) 147826.472 384.4820831 98.85% Total 149552.651 386.720378 100.00%
USL LSL Precision to Tolerance Ratio Precision to Total Ratio 0.107435 Resolution 13.0
-5
0
5
10
15
20
25
0 200 400 600
time
to T
-dow
n (h
ours
)
VDK (ppb)
VDK reduction slope > generation 1
yeast pitched measurement analysis
Cell count Slurry volume (spin down) Source Variance Standard Deviation % ContributionTotal Measurement (Gage) 37419.18 193.44 7.91% Repeatability 31346.80 177.05 6.63% Reproducibility 6072.38 77.93 1.28% Operator 6072.38 77.93 1.28% Oper * Part InteractionProduct (Part-to-Part) 435537.50 659.95 92.09%Total 472956.67 687.72 100.00%
USL 2000.00LSL 0.00Precision to Tolerance Ratio 0.58Precision to Total Ratio 0.28Resolution 4.8
Source Variance Standard Deviation % ContributionTotal Measurement (Gage) 5.32 2.31 0.86% Repeatability 5.00 2.24 0.81% Reproducibility 0.32 0.57 0.05% Operator 0.00 0.00 0.00% Oper * Part Interaction 0.32 0.57 0.05%Product (Part-to-Part) 612.69 24.75 99.14%Total 618.01 24.86 100.00%
USLLSLPrecision to Tolerance RatioPrecision to Total Ratio 0.09Resolution 15.1
3x increase in resolution
Viability MBà resolution 1 L
Base line measurement
std. dev 720hl
average 720hl std. dev. 1440 hl
average 1440hl
0
10
20
30
40
50
60
70
80
90
100
F time KO to 24C 18to24C at 24C
14.6 14.8
6.5 9.7
95.6
54.7
29.2
38.3
10.5 7.5
5.1 5
91.6
48.2
23.4
39.2 std. dev 720hl
average 720hl
std. dev. 1440 hl
average 1440hl
Ferm
enta
tion
time
(hou
rs)
Trivial Many à Vital Few Variables: • Yeast:
amount, vitality, viability, generation, time of addition,…
• Fermenter conditions: Number of brews, temperature tank, KO temperature, outside temperature,….
• Fermentation heat up rate • Nutrients:
Sugar profile, aeration, minerals, vitamins, FAN,…
Variables yeast’s past: • Yeast storage time and
temperature, • length of previous
fermentation, • fermentation cool down
rate • Fermentation heat up rate • Anomalies
Is this yeast vitality? DOE Pro
Sigmazone®
Modeling
Voilà Voilà
Only variables within current fermentation
fermentation:variable abbreviation unit variable abbreviation unityeast pitch volume Vol/br hectoliter length of fermentation pFVt hrcell count cells cells/ml time until VDK down pVDKt hrviability viab % yeast storage time Ystime hrgeneration genKO temp. to 16C t to 16C hr KO temp. to 16C pt to 16 hr16C to 18C 16 to 18 hr 16C to 18C p16 to 18 hr
exothermic 18C to 20 C 18 to 20 hr 18C to 20 C p18 to 20 hrtimes KO temp. to 20C t to 20 hr KO temp. to 20C pt to 20 hr
T in fermenter KO1 T1st Cstart of T in fermenter KO2 T2nd Cfermentation T tank T tank Cconditions number of brews #brews
yeast related
previous fermentationcurrent fermentation
0.0 0.5 1.0 1.5 2.0
vol/br
T2nd
t to 16
16 to 18
T1st
cells
18 to 20
Ttank
t to 20
gen
viab
#br
within fermentations
Compilation from modeling historical data with DOE Pro® from Sigmazone ®
Variable included from previous fermentations
Individual variables less important than their interactions!
0.0 0.5 1.0 1.5 2.0
T2nd16 to 18pt to 20pVDKtt to 20
T1st18 to 20
t to 16Y vol/brYsttime
pFVtcellspT5
pT10pT3
p18 to20generation
T tank#brews
with previous fermentation
fermentation:variable abbreviation unit variable abbreviation unityeast pitch volume Vol/br hectoliter length of fermentation pFVt hrcell count cells cells/ml time until VDK down pVDKt hrviability viab % yeast storage time Ystime hrgeneration genKO temp. to 16C t to 16C hr KO temp. to 16C pt to 16 hr16C to 18C 16 to 18 hr 16C to 18C p16 to 18 hr
exothermic 18C to 20 C 18 to 20 hr 18C to 20 C p18 to 20 hrtimes KO temp. to 20C t to 20 hr KO temp. to 20C pt to 20 hr
T in fermenter KO1 T1st Cstart of T in fermenter KO2 T2nd Cfermentation T tank T tank Cconditions number of brews #brews
yeast related
previous fermentationcurrent fermentation
Time from cooling on until certain beer T reached
Single secondary parameter: Green beer cool down time variation
cooling rate, time (hrs) to reach 10C
UCL=17.387
LCL=4.22
CEN=10.804UCL=12.158
LCL=7.602
CEN=9.88UCL=11.373
LCL=7.9854
CEN=9.6792
0
2
4
6
8
10
12
14
16
18
20
1844
218
46418
47918
51118
53718
558
1858
318
62218
63918
67018
69518
72318
746
1877
818
79818
82418
85618
87818
90218
928
1895
518
99019
02819
04619
06419
10519
141
1916
219
19919
23219
25319
27919
30019
320
1935
019
36519
40019
46919
56819
59819
666
Different group of fermenters used
Design of Experiments (DOE)
• Two variables – T of knock out
• (1)-0.5&+1C, (2)+1&+2C – Yeast pitching rate.
• (1)+/- 16.7%, (2)-0,-16.7% – 5 repetitions randomized – Same tank size/6 brews
• Confounding variables: – Yeast harvests from tests
• Time bound: outside T 1.2
1.4
1.6
1.8
2
2.2
2.4
13 13.5 14 14.5 15 15.5 16 16.5 pi
tchi
ng r
ate
KO T (°C)
DOE 1
DOE 2
DOE 1: Data analysis
y = -0.1186x + 4685.5 R² = 0.40222
10.0
12.0
14.0
16.0
18.0
20.0
22.0
14.0
14.5
15.0
15.5
16.0
16.5
8/16 8/26 9/5 9/15 9/25 10/5 10/15
Tem
pera
ture
in F
V lo
wer
pro
be a
fter
KO
3 (C
elsi
us)
all
LL
LH
HL
HH
FV T
Linear (all)
Linear (FV T)
65
70
75
80
85
90
8/21 8/31 9/10 9/20 9/30 10/10
ferm
enta
tion
time
(hr)
all
LL
LH
HL
HH
Linear (all)
DOE 1 results: closer look
1.2
1.4
1.6
1.8
2
2.2
2.4
13 13.5 14 14.5 15 15.5 16 16.5 pi
tchi
ng r
ate
KO T (°C)
DOE 1
DOE 2
Shorter fermentation time
Less variation 13.5 13.83333333
14.16666667 14.5
14.83333333
0.0
1.0
2.0
3.0
4.0
5.0
6.0
1.5 1.6 1.6 1.7 1.8 1.8 1.9 2.0 2.0 2.1
KO temperature ( C)
stan
dard
dev
iatio
n (h
ours
)
pitch
Fermentation time S-hat Surface Plot KO T vs pitch
5.0-6.0 4.0-5.0 3.0-4.0 2.0-3.0 1.0-2.0 0.0-1.0
What did change?
• KO T à T in fermenter (yeast T, PID) • Increased KO T for 1 generation • Ferment in certain set of fermenters • Stopped using viability through Methylene
Blue à heat up time • Spin down lab centrifuge test for pitch • Pitch slightly less • KO T used for seasonal fluctuations
How will you hold the gain? Control Charts / QA group
UCL=90.315
LCL=60.926
CEN=75.62
UCL=91.826
LCL=64.816
CEN=78.321
USL=95
LSL=70
50
55
60
65
70
75
80
85
90
95
100
3000
8 30
023
3003
7 30
047
3004
6 30
071
3008
1 30
098
3010
4 30
125
3014
0 30
158
3017
2 30
178
3019
2 30
198
3021
7 30
226
3024
2 30
256
3026
7 30
280
3028
9 30
309
3031
5 30
324
3033
1 30
342
3035
3 30
367
3040
6 30
413
3042
0 30
434
3046
1 30
482
3048
8 30
545
3055
4 30
582
3061
1 30
628
3064
6 30
655
3067
3 30
697
3072
3 30
739
3076
4 30
772
3079
8 30
809
3082
4 30
833
3084
5 30
857
3086
8 30
877
3087
7 30
892
3091
9 30
922
3095
9 31
014
3103
2 30
164
FT fermentation time (hours)
What did we gain?
• $8 M in 2007 à $3.7 M in 2010 – Queuing and fermentation variation!
• Stopped inaccurate measurement: – Methylene Blue – Cell count
• Automated measurements through MES – Heat up rate of fermenters:
• Early warning that fermentation is “off” • Actions: no harvest source, blending, KO T
change
Reducing Fermentation Variation
Peter Bouckaert New Belgium Brewing Co