The geometry of experience curves can be seen on youtube.

http://www.youtube.com/watch?v=3EW1DnJg1qY

A: For an Economy, like say the U.S. where we have variables Employed Men, Price,Volume,Wages,Investment,Scale of Operation, Production Process Mean Dimensions of Radius R with Number of space dimensions being used = m, Dollars, A Gross National Product, Specific Products and the change in time of these variables then:

For a change in Production of a Product with time the Behavior of Men is to Minimize their time/Per unit of Volume of the product, i.e. that ratio (dMen/dRadius with time)=0.

Using reduced values, or indexed values of these variables based on a given year for GNP, Price, Volume, etc. and the Fact that Dollars=Price*Volume=Wages*Men, and Volume=Radius^m gives a Slope for the Experienc Curve=(1/m-1)*(1-Exp(-k*t)) wherE k=growth rate constant for the product and t=time (usually years) .

Because “All Men involved minimize their time per unit of space dimension, with time”.  Thus this Law of Nature is claimed as Garnett’s Minimization Law

(more)

Inherent in the foregoing is also the fact that Investment=Scale^(1/m).

The price of the flu is death of a fraction (shown below) of those infected.  The volume is those that have been infected.  The experience curve of the h1n1 flu follows:

recent-us-deaths

Which has slope of 2.

The number of dimensions m, normally being used is: Slope=1/m-1, hence in this case 1/m=2+1=3.

However, we are not infecting the virus it is the other way around.  Hence from the viruses point of view the number of dimensions is 1/m not m, and therefore=3.  It is a volumetric production rate.  Presumably the graph of deaths vs. invections will continue until the number of uninfected people is beginning to diminish, somewhere over half of the population.

The death rate per thousand infected people (the “fraction”) has been rising but recently has been diminishingas shown in:

recent-us-deaths1

Once you accept that the investment for a tank or reactor or container scales up with the 1/3rd power of volume then the scale up with pressure naturally follows.

If the rate of reaction of a chemical reaction is proportional to pressure (P) then raising pressure will lower the volume required at the same time that it will raise the thickness of the walls to contain that higher pressure.  It follows that Investment will be:

Investment=k*(1/P)^(1/3)*P

Investment =k*P^(2/3)

If the reaction is proportional to the pressure squared (2 species->one species)

Investment=k*(P)^(1-2/3)=k*(P)^(1/3)

For a 3rd order pressure dependency on the reaction rate the Investment is independent of pressure, i.e.

Investment=k*P^(1-3/3)=k

For orders higher than 3 other things like heat transfer to/from the reactants (i.e. area or order 2),  or limiting conversion to limit temperature rise, or use of diluents to control temperature rise,  may become dominant, limiting the pressure effect between P^0 to P^(1/3).

By my memory the investment for polyethylene plants scale as the 1/4th power of Pressure.

The foregoing is applicable for a fixed production rate.  However if one is in the position of building a plant where the justification of the capital is a key element then one likes to consider investment per annual pound.  Larger plants tend to give lower investments/annual lb.  Since production rate is proportional to P raised to the power of 1,2, or 3 preceding the investment per annual lb will be proportional to Pressure raised to the power of -1/3, -5/6, -6/3=-2.  That being the case plants tend to be at the highest pressure and largest scale.

Is the Learning Curve the cause for the Experience Curve?

Bruce Henderson in his “Perspective on Experience” alluded to learning being responsible for the experience Curve being the way that it is. However, he never did quantify the relationship between price and cumulative volume much less their relationship to learning. Henderson’s lack of understanding of the cause and effect relationships is exemplified by the many assertions of not knowing why the curves were as they were. Michael Rothschild in his Bionomics: Economy as Ecosystem asserts Henderson as writing:

“The experience curve phenomenon is as real as gravity. . . . [Its] effect can be observed and measured in any business, any industry, any cost element, anywhere. . . . The reason for the experience curve effect are not particularly important. The important fact is that the experience curve is a universally observable phenomenon.”

Clearly Henderson did not know and did not want to know the reasons behind experience curves as long as his experience curve phenomenon continued to be acknowledged.

See link for more:

rats_motivation

Why do scale up rules of thumb work and why is the historic six tenths power six tenths?

The cause and effect relationship is shown in this link (publication)

scale

I was talking to Gregory S. Patince recently about Research being an Investment, or not.  My view of course is that it can be if managed to be an investment.  Just how that is so is included in the following documents.  The criteria of what is required is defined.

research-as-investment-power-point1

research-as-investment-manuscript

These documents are part of the continuing dialog about what Experience Curves are and how they can be extended beyond historic data.

The equation for the slope of an experience curve is (1/m-1)*(1-exp(-kg*t)) or
(SF-1)*(1-exp(-kg*t)), where: t is time typically years, kg is the exponential growth rate

constant, SF is scale factor from the slope of an xy plot of scale of the facility vs investment

 in constant dollars, m is the number of dimensions of the production facility eg. 1 for linear

 like a pipe line, 2 for an area facility like a plate and frame filter press, and 3 for a volumetric

 facility like a tank. For a mature product slope becomes (1/m-1) or (SF-1).

(more…)

The post titled “Money” shows the GNP or money=k*(employed)^4 and volume=k*(employed)^2.16 and

price=k*(employed)^1.84. In the context of these posts the number of dimensions of the production

facilities, m=2.16 and the price exponent is (4-m). Differenting by parts results is the fraction of the change

in GNP that is real growth is m/4 and the part that is inflation is (4-m)/4. The value of m therefore determines

the fraction that is inflation and real growth. The value of m determine the fraction that is real growth, namely;

m=1 results in 25% real growth,

m=2 results in 50% real growth

m=3 results in 75% real growth.

Obama’s and others “put people to work” with bail-out government money will not help the economy much if

the number of dimensions of  those activities is low (e.g. a shovel in hand ditch digger) vs. high

(like building a petrochemical plant).

the-geometry-of-experience-curves-aiche5-cut

Defines what an experience curve really is and why.