1. ABSTRACT
Everyone makes the best extracts. But are they optimized? And what does “the best” mean; does
it mean high yield, high quality, or both? And what is quality anyway?
We studied our rosin to optimize production, finding the ideal balance between yield and consumer
rating. Our methods allow for a aromatic, designer quality extract, or a high yielding oil.
All of this was achieved in an efficient manner through the use of design of experiment (DoE)
techniques.1
INTRODUCTION
Experimental design is a structured approach to investigate reaction space to improve certain
responses, e.g. yield or quality. The process of experimental design, or DoE, can be broken down
into distinct steps.
1. Define the problem
• Extracting Cannabis with a 140psi Longs Peak Rosin Press
2. Determine the response
• % Yield
• Subjective Quality
3. Determine influential factors
• Temperature, pressure, time, actuation speed
4. Additional things to keep in mind
• Comparison: Ensure that every experiment is done under the same conditions.
• Blocking: Do the experiments in one sweep to mitigate variation in starting material or
instrumentation.
• Orthogonality: Choose factors that complement each other, not overlay.
5. Calculating response surface
• A Taylor expansion describes the response surface:
• Matrix algebra solves the Taylor expansion:
Y=bX; b=(X’X)-1X’Y; a=bX
DISCUSSION/CONCLUSION
We conclude that DoE is a valuable tool to optimize rosin production quickly.
Temperature and pressure have the largest influence on yield and quality. Specifically, temperature
correlates positively with yield, but negatively with quality.
Rosin was elevated using this DoE, with a minimal waste of material, to a high-yielding and
high-quality offering, consistently produced on standardized SOPs.
RESULTS
Multivariate Optimizations show:
• Product Quality(Fig. 1) and Yield(Fig. 2)
These two opposing response surfaces inform the optimal conditions.
• Temperature has a direct influence on quality and extraction efficiency.
METHODOLOGY
DoE principles were used to study the effects that pressure, temperature,
actuation, and time have on cannabis oil yield and quality. A homogenized
mixture of a single cultivar was used with a water activity of 0.58 or ERH of
44%. Ambient conditions were kept at a steady 64˚F. The oil is filtered
through 115µ screens at ranges from 15 to 120psi and 150 to 300˚F for 60
to 240 seconds. A five-factor uniform shell design was used with thirty one
distinct factor combinations.
All fractions were analyzed for total yield, and overall quality with a
designation between smoking or infuse-able grade. Data mining of the test
results were done using Microsoft Excel. Calculations of second order
Taylor expansion and response surfaces (Figs. 1-2) were done using
Mathematica.2
References:
1) Fisher, R.A., The Design of Experiments, Second Edition. 1937: Oliver and Boyd
2) Stansbury, W.F., Chemometrics and Intelligent Laboratory Systems. 1997, p. 199
Quantity
Fig.1 Fig. 2
Flower weight
Temperature
Actuating
Pressure
Time
10 - 20 g
150 - 300 ˚F
2 - 10 turns
15 - 120 psi
60 - 240 sec
Quality
Multivariate Optimization of Rosin Extraction
Travis Kennedy, Blake Grauerholz
OutCo, San Diego, CA
Multivariate Optimization of Rosin Extraction
Travis Kennedy, Blake Grauerholz
OutCo, San Diego, CA