Massachusetts Institute of Technology Manufacturing Science/Engineering of Solar Cells Assignment - Hypothesis Testing of New Wafer Suppliers
Massachusetts Institute of Technology Manufacturing Science/Engineering of Solar Cells Assignment - Hypothesis Testing of New Wafer Suppliers
Massachusetts Institute of Technology Manufacturing Science/Engineering of Solar Cells Assignment - Hypothesis Testing of New Wafer Suppliers
Your Manufacturing Engineering team is evaluating multiple second-source silicon wafer suppliers in order to
diversify Virtual Solar Inc.’s wafer supply chain. Having multiple qualified suppliers both de-risks potential supply
and-logistics issues and puts pressure on our suppliers to lower their prices. Eight new suppliers are under
consideration – of principle concern are (1) wafer price, (2) wafer power performance, and ultimately, (3) $/watt cost
changes in the product. Of course, each supplier promises better cost, performance, quality, reliability, yield …
Pricing data is provided courtesy of VSI’s Purchasing Department, which has been in preliminary negotiation with
each supplier. Purchasing also acquired evaluation samples from each of the new wafer suppliers – 2000 wafers,
which is enough to process 20 “lots” of 100 wafers in the fab. The table below summarizes the preliminary pricing
data for each supplier.
Your team will evaluate one wafer type on your production line, and we will compare notes between teams as a class.
To complete the evaluation, your team needs power performance data. However, with only a limited supply of
evaluation wafers, “luck of the draw” and “common cause variance” will surely introduce uncertainty into your
evaluation. So, you will use a statistical hypothesis test approach to quantify and communicate this uncertainty. In
particular, your team will use a “Two-tailed, 95% Confidence, Hypothesized Mean, Z-hypothesis Test About Lot-
Mean Maximum Power (Pmp).” This type of test has been stipulated by the Director of Manufacturing Engineering.
Her rationale is
1
Assignment Objectives:
1. Use the VMES Engineering > Data Query and Export feature to query lot-mean Pmp data from 4 initial
engineering lots pertaining to your assigned wafer type (type A-H). These lots that have already been
fabricated by Jeff and are ready for evaluation. See the Helpful Tips section at the end of this handout.
2. Use statistical hypothesis testing to evaluate the data pertaining to your assigned wafer type, against the
current baseline (see the table above for baseline data). Start with data from the initial 4 lots already prepared
in the VMES. Report your findings (Zstat, p-value and conclusion) to the class.
3. Use the VMES Engineering > Action Planner > Eng Lots feature to create 16 additional engineering lots in
the VMES. There’s an easy shortcut, Jeff will show you in class. When the VMES updates again (around late
o’clock in the night), these lots will be fully fabricated and ready for data query and further analysis.
4. Repeat the statistical hypothesis testing to evaluate 20 lots of data. Report your findings (Zstat, p-value and
conclusion) to the class.
5. Compute the potential performance cost change (delta $/watt against the baseline performance cost) for your
wafer type.
What to Submit:
1. One report per team, not more than 1 page in length.
2. A concise table of your hypothesis test results for the 4-sample and 20-sample evaluations: Zstat, p-value,
conclusion. Be sure to state your conclusion properly (eg. “Fail to Reject with 95% confidence and a p-value
of 0.345”)
3. Compute the delta in performance cost ($/watt to the 100th of a cent) compared to the baseline case. Make the
best possible estimate from your available data. For example, report a savings as –$0.0123/watt and an
increase as +$0.0098/watt.
4. Recommend whether your assigned supplier should be considered as a supplier to Virtual Solar Inc. Make a
concise recommendation with support from your data and analyses.
5. Consider the results from the rest of the class. Recommend the best supplier to consider for Virtual Solar Inc.
Make a concise recommendation with support from your data and analyses.
2
Helpful Tips
1. Getting the Virtual Manufacturing Execution System (VMES) Software
a. The VMES software is free. Download it from https://sites.google.com/site/virtualsolarcellsinc/
b. After downloading, run the installer. NB: the software runs well on Windows PC, but may have difficulty
on Mac emulators.
2. Logging Into the Virtual Cell Factory (VCF) using the VMES Software
a. The Virtual Cell Factory is in the cloud, you access it using the VMES software.
b. Select Account > Login and enter your team’s Login Name and Password. Login Name and Password are
the same for the MIT class, depending on your assigned line number (“mitline1”/”mitline1”,
“mitline2”/”mitline2”, etc).
c. Click Connect, then Select, then Login (your class VCF is “University/MIT VCF 2013”).
3. Setting up a Query
a. Select Engineering>Data Query and Export, and then fill up the appropriate query parameters on the
form. See the illustration below.
b. Click the Query button. From the resulting table, you can use ctrl-c to copy the pertinent data.
3
MIT OpenCourseWare
http://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.