JMP Solid Product Cooling

Solids are a convenient form for consumers
to apply a product onto their skin.

Deodorants, antiperspirants, sunscreens
and bug repellants are examples. A
common way to formulate and produce
them is with an active ingredient
suspended in a molten liquid containing a
molten gellant ingredient. This is then
cooled to close to the gellant’s
solidification point and poured (aka
“filled”) into the packgage at high speed.
The resulting products are then cooled at a
controlled rate in a convection, cooling
tunnel having relatively long residence
time of 30+ minutes. In addition to
solidifying the product so that the active
ingredient does not settle, the cooling can
affect the final hardness of the product. That is problematic if not controlled. If the product is too soft, it will smear on
the consumer’s skin. If it is too hard, not enough product will be applied, and it will not work as desired.
The data set consumer_solid.jmp contains cooling data from

hypothetical lab and plant scale cooling experiments with such
products. The temperature data are the interior temperature of the
product measured by inserting a temperature probe into the center
at various times while products are cooled down from the filling
temperature. These measurements are taken at the furthest,
interior location away from the package walls and other package
parts such as the screw elevator used to dispene the product.
Exercises Using JMP® software:

1. Plot the raw temperature versus time data using color-coding for type of process
2. Add a dimensionless temperature column, Theta (q), to non-dimensionalize temperature to a 0 to 1 scale
ranging from the the initial, filling temperature (q = 1) to the final, ambient temperature (q = 0)
3. Make it possible to linearize the data by adding a - ln(q) column
4. Plot - ln(q) versus time and fit straight lines by type of process (exclude initial data points where the product’s
core temperature has not yet started to react to the cooling process)
5. Calculate each process’ characteristic cooling time constant, t, based on the limiting case theory below.
JMP Step-by-step Guide for Exercises

The consumer_solid.jmp data contain Temperature versus time data for two cooling processes –Lab and Plant. Based on
the Theory section below, to quantify/characterize the Lab and Plant processes’ cooling, we should convert temperature
to a dimensionless temperature and plot the negative natural log of it. If there is a good, linear fit with that plot, we can
then calculate a characteristic time constant for each process from the plot’s slope.
To explore this in JMP:
1. Open the data and make a plot (Analyze / Fit Y by X) of Temperature (Y) by time (x)
2. Use Rows / Color or Mark by Column to color code the data by Process
T versus Time
Fit Y by X Plot
• Which process has faster cooling? Lab or Plant?

• What is the initial behavior of the core temperature? Does this make sense physically based on the product or
process?
3. Add a calculated dimensionless temperature, Theta, column

a. For these data, the initial temperature, T0 is 50 C, and the ambient temperature is 23.0 C. Realistically,
these might differ between lab and plant, but we keep them the same here to simplify.
b. To add a new column to the JMP data table, either double-click in the data table’s column header region
or choose Cols / New Column menu to add a new blank column
c. In its Column Info (double click on header or choose Cols / Col Info menu), set its name to “Theta” and
select “Formula” under Column Properties as shown
d. To enter the formula (T – Ta)/(T0 - Ta), double-click on the T_degC column name to use that column’s
values. Type a minus sign, 23 and Enter. This creates the numerator
e. Once the numerator is entered, click on the bigger rectangle surrounding the equation to select it (blue
highlight) and type a slash “/” or the divide-by operator
f. Type “50.0”, Enter, “23.0”, Enter to input the denominator
è
è è
g. Click OK twice to apply the formula. Notice that, because Theta is a calculated column, JMP prohibits
manual changes to its values
4. Add a “-ln(Theta)” column to calculate the negtive log of dimensionless Temperature, Theta.
a. Add another new column and select Formula / Edit Formula in its Col Info as before
b. The “ln” function is under the Transcendental grouping in the formula entry dialog
c. Use the “+/-“ button at the top of the formula entry to flip the equation’s sign to minus
d. Remember to “select the whole equation by clicking on the outer shaded rectangle (blue highlight) so
that you end up with - ln(Theta) and not ln(-Theta)
5. Plot -ln(Theta) versus time. Based on the theory, we expect this to be linear for each process –except for the
initial times where the center of the product has not yet started to cool.
6. Before doing the linear fit, we need to do two actions:
a. Lasso and exclude points in the initial “induction” period where the data are not linear. It is a judgement
call how many to exclude, but the cooling appears to reach steady state below q ≈ 0.97
b. Tell JMP that we want to use Process as a Group By variable. This selection is under the plot’s red
triangle menu as shown1
1
Note that an alternate way to “Group” is to assign the Process variable to the Fit Y by X dialog’s “By” listbox when making the plot.
This will result in separate plots for Lab and Process, and you can then perform the linear fit for each.
-ln(Theta) versus time
Fit Y by X Plot
7. Fit lines to the data by choosing Fit Line from the Plot’s red triangle menu. The result in a good linear fit: R2 =
0.995 for Lab and 0.992 for Plant
Fit Line with “Process” as

Group By variable
8. Calculate the time constants, tc as 1/slope for each cooling process. Do they make sense? Since this is a first-
order equation, cooling should be approximately complete (e.g. to ambient) in 3tc
9. What do you suggest changing about the lab process to reduce tc to more closely match the Plant cooling
behavior?
Background Theory
The raw, cooling data are not linear. We could just fit a non-linear model to the data, but that will not simplify things or
give us a single number to use to quantify what each process is doing.
Two, limiting cases help guide on how to transform the data based on heat transfer theory2:
1. What if the product and its packaging have no (or negligible) conduction resistance relative to the
convection rate? This would be the case if the product and package were made of a high conduction solid
such as copper, silver or gold.
• In this case, all heat removed by the cooling instantly shows up at the center of the product in the form
of reduced temperature.
• There is a first-order, ordinary differential equation solution for that unsteady state problem
• The time constant, tC, is a function of the solid’s mass, heat capacity, surface area and the convection
heat transfer coefficient, so it is a constant for a given solid in a given cooling environment
t
- ln q =
tC
Where:
q = (T – Ta) / (T0 – Ta)

tc = mCp / hA
h convection heat transfer coefficient

A effective surface area of the solid
T time-dependent temperature of the solid
T0 the solid’s initial temperature
Ta ambient temperature
M mass of the solid
Cp heat capacity of the solid
2. What if the product and package are a homogeneous sphere, and the convection to conduction ratio (Biot
number) is relatively high?
• There is an infinite series solution for time-dependent conduction in an ideal solid
• The infinite series solution captures initial behavior where there is a time lag before heat transferred
to the outside of the solid shows up as temperature change at the center. The first term captures
the temperature change at the center following this induction period
- ln q 0 = - lnC1 + [z 1
2
]
k rC pr 2 t
• The bracketed quantity is a constant based on the solid’s properties and the cooling process (e.g.
convection rate)
• C1 and z1 are functions of Biot number aka ratio of convection rate and properties of the solid
• Additional infinite series terms serve to “round off” dog leg but don’t change the fundamental shape
of the –ln(q) versus time graph.
These limiting cases both guide to plotting -ln(q) versus time and treating the inverse slope as a characteristic time
constant for the product + process combination. The limiting cases suggest using a dimensionless temperature.
2
D.P. DeWitt and F. P. Incropera, Introduction to Heat Transfer (3rd Edition), John Wiley and Sons, New York, 1996, pp. 225-236

JMP Solid Product Cooling

Uploaded by

Copyright:

Available Formats

JMP Solid Product Cooling

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

JMP Solid Product Cooling

Uploaded by

Copyright:

Available Formats

Solids are a convenient form for consumers

to apply a product onto their skin.

The data set consumer_solid.jmp contains cooling data from

Exercises Using JMP® software:

JMP Step-by-step Guide for Exercises

To explore this in JMP:

• Which process has faster cooling? Lab or Plant?

3. Add a calculated dimensionless temperature, Theta, column

Fit Line with “Process” as

q = (T – Ta) / (T0 – Ta)

h convection heat transfer coefficient

You might also like