Attached are the slides presented in the Sales Track at Kick-Off 2019.
Please use this thread to share success stories, war stories and practical examples of using FDE to drive opportunities with our customers.
============
Product Session – Functional Data Explorer
Damian Carville, @DamianCarville – Regional Sales Manager, Americas
Chris Gotwalt, @chris_gotwalt1 – JMP Director of Statistical R&D
Product knowledge
How much is enough?
What is JMP?
A tool to help you better understand your process
JMP is (primarily) designed for Scientists & Engineers wishing to:
Improve innovation in R&D
Reduce waste/cost in manufacturing/service operations
Functional Data Explorer - session objectives
What problems are addressed by Functional Data Explorer?
Improve innovation
Reduce waste/cost in manufacturing
How do customers approach these issues today?
Competition
How do I find targets for Functional Data Explorer?
What is ‘functional’ data?
A lot of data is really functional in form:
- Time series data
- Sensor streams from manufacturing processes
- Spectra
- Measurements taken over a range of temperatures
We approach these issues from 2 angles:
Functional Design of Experiments
Design TO A PROFILE
Use enhanced DOE to predict a response function (potency over time, force across distance)
How can I make something behave according to a particular profile?
Functional AI
Utilize the ‘curve data’ to better understand what is happening
Using functional data (sensor streams) to predict something (yield, production faults)
How can I predict (or understand) something using functional data?
Functional DOE – An Example
- Active ingredient particles for pharmaceutical formulation are milled
- The particles need to be within 70nm-85nm
Functional DOE - Applications
There are many applications of F-DOEs:
- Dissolution profiles for tablets/pills
- Formulation to match a particular spectra
- Product stability/shelf life
- Pharmacokinetics
- Shear/viscosity profiles of engine oils
- Force/distance profiles (e.g., safety belts)
- Surface tension over time (e.g., consumer dish soap)
Functional AI Example - bonding
- Automotive gyroscope uses a semiconductor manufacturing process with a step where glass is bonded to the devices (anodic bonding).
- The anodic bonding step is problematic: it destroys 12% of the wafers.
- The wafers cannot be tested for several weeks, after the rest of the production process steps are applied.
- The bonding tool has several sensors that take real-time measurements of charge, vacuum, flow, voltage, & piston force.
- Can we get early prediction of damaged wafers and screen them out?
Functional AI - Applications
- F-AI has already had successes:
- Predicting pharmaceutical batch yield using sensor streams taking measurements in the tank
- Early fault detection in semiconductor manufacturing using sensor streams
- Extract FPCs from multiple streams, and fit neural networks or generalized regression models to the parameter of interest…
Functional Data Explorer - session objectives
- What problems are addressed by Functional Data Explorer?
- Improve innovation
- Reduce waste/cost in manufacturing
- How do customers approach these issues today?
- Competition
- Umetrics SIMCA-P
- How do I find targets for Functional Data Explorer?
