Experience Curves: How They Work and What They Foretell About Lithium Ion Storage Prices

Elon Musk’s announcement of the Tesla Giga Factory garnered a great deal of attention.  Among other things he mentioned that by building this large factory they were sure that the cost of Lithium Ion batteries would come down.  How could he be so sure?  What gives Elon Musk the confidence to project out half a decade and more.  Is it a deep understanding of all of the nuances of battery technology?  His team has likely done analysis.  But the more likely explanation for his confidence is that he has looked closely at the experience curve for lithium ion batteries and done a simple extrapolation.  What is an experience curve and how can it be used to forecast an entire industry?  It is simply the mathematical relationship that describes cost reduction of manufactured product as a function of total cumulative production.

As manufacturers gain experience in mass producing a product they improve their processes and efficiency.  They figure out how to save time, reduce materials, reduce rejects, improve quality and so on.  It has been found that, in general, as more units of a given product are manufactured the cost to manufacture them declines in a predictable way.  This can be seen graphically in what is called an experience curve, also known as the improvement curve.  See Figure 3.

A familiar example of a product that declined in price in response to growth in sales is the flat panel computer monitor.  12 years ago my company purchased a 15″ flat screen computer monitor for $500.  7 years later I purchased a 19″ monitor for $200 and today a quick Google search shows name brand 24″ monitors for under $150.

Solar Module Experience (Improvement) Curves

Solar Module Experience (Improvement) Curves

Figure 1  Experience curves for solar modules.  Click graph to enlarge

A less familiar example, that has been in the news lately, is the rapid decline in cost of solar photovoltaic (PV) modules, see figure 1 above.  It makes sense to most people that if you produce lots of something its cost will decline.  However people are often surprised to learn that it continues no matter how many are produced.  It has also been found that this cost reduction with experience can be described mathematically.  This allows people to make reliable predictions as to the cost at some time in the future.  Below is a graph showing the experience curves for both silicon and CdTl solar modules.

Experience curves have been found to apply universally to manufactured products.  In the mid 1930’s T.P. Wright noticed this declining cost phenomenon while managing the production of airplanes.  Wright was not the first to notice that cost reduction resulted from volume production.  What was new was that he showed that cost reduction was mathematically predictable.  Once enough units were manufactured to deduce the doubling constant, costs could be extrapolated mathematically with surprising accuracy.  It can also be seen in production of cars as in the Model T Ford, see figure 2

Experience Curve William J. Abernathy and Kenneth WayneFigure 2:  Price of Ford Model T 1909 – 1923 (in constant 1958 US$) versus cumulative units produced. 15% doubling constant. Source: W.J. Abernathy and K.Wayne. Limits of the learning curve. Harvard Business Review, September – October 1974.

The basic math of experience curves is accessible to the layman.  Simply stated, for every cumulative doubling of production the cost reduces by a fixed percentage.  The fixed percentage is referred to as the doubling constant.  Cumulative production refers to the total number of the product ever produced.  In practice let’s see what that means.

Experience CurveFigure 3:  Example of cumulative production and experience curves.  The far right graph is the same data as the middle graph.  The difference is that it is plotted on logarithmic axes.

Let’s say you invent something and after producing 1000 of them you ask your staff to tell you how much it will cost to produce the next one.  Suppose they find that the cost will be $100 dollars to produce unit number 1001.  Further, they inform you that they studied the cost of your devices since they first were produced and found the doubling constant to be 10%.  You already saw that the total produced to date was 1000.  So a cumulative doubling would occur when production reached 2000 units.  Applying the 10% doubling constant you find that the cost will reduce by 10% to $90.  How many more need to be produced to attain the next 10% cost reduction?  The answer is not 3000, it is 4000.  Each drop in price requires a doubling of the total number produced.  So after 4000 units, the 4001th unit will cost $90 x 10% = $81.  And so on, See Figure 3.

Both of the graphs in figure 3 are of the same data.  The difference is how the scales are laid out.  In the first graph the numbers are spaced evenly.  On the second graph the numbers on each axis are spaced logarithmically.  This is called a log-log plot and is used to spot exponential trends.  An exponential function plotted on a log-log plot will appear linear.  Experience curves are theoretically exponential functions.  As you can see from the experience curves in Figures 1 and 3, lithium ion battery and solar photovoltaic module production both lay nicely on the line.  The silicon PV modules have a few excursions from the line which are explained in the Maycock paper cited below.

The usefulness of experience curves extend beyond a single factory or even a single company.  Experience curves apply across entire industries.The cost reduction of solar modules has been tracked for decades.  The doubling constant has been found to be 17%.  In 2002 I was considering taking some business risks related to the solar industry.  I wanted to convince myself that that was a good idea.

In this 2002 paper, I used data compiled by Paul Maycock to predict that, if the PV industry grew at 40% annually, we would see $1.00 / Watt solar modules in 2011.  Even optimists were skeptical at the time and few took these numbers seriously.  As it turned out the industry did grow at that rate and we did hit $1.00 / Watt in 2011, right on schedule.

Since the collapse of solar photovoltaic pricing in the last few years, many feel that the remaining hindrance to high penetration of solar photovoltaics is lack of storage.  However, f one believes the improvement curves for Li-ion batteries then electric vehicles will hit the tipping point between 2018 and 2020.  Hyper growth of EV sales will further accelerate demand for lithium ion batteries.   If this growth continues the experience curve suggests that lithium ion batteries may well be the storage solution that allows near total decarbonization of the utility grid.

Bloomberg developed a graphic showing the experience curve for EV batteries and lithium ion batteries used in consumer electronics such as laptop computers.  They show them on separate curves.  However, EV batteries are lithium ion batteries.  It appears that the high cost of EV batteries may have more to do with developing low cost ways to package the batteries for safe use in cars.  That technology is being rapidly developed and packaging costs are dropping rapidly.  It appears that the EV battery pricing will soon track the consumer electronic battery pricing, figure 4.

Recent press announced that Nissan has introduced a low cost replacement battery for the Leaf.  Owners of older Leaf’s whose batteries degrade, can now replace them for $6,500.  This amounts to $270 per kwhr and puts EV batteries into the cost range that many anticipate will make EVs much more attractive to the general public.  In figure 3 below I have added the recent Nissan announcement to show the extraordinary progress already made in reducing the cost of lithium ion batteries.

Li-Ion battery Improvement CurveFigure 4:  Experience curve for lithium ion batteries showing the rapid cost declines and the movement of EV battery costs to the consumer Li-ion curve.  Click Graphic to enlarge.

This is like deja vu all over again.  Just as with PV, people are saying that lithium ion batteries are simply too expensive and cannot reach acceptable price levels in the foreseeable future.  However the experience curves tell a different story, an exciting story and a story that offers hope that the Achilles heel of renewables, storage, may soon be no more.

Renewables related experience curves for various products.

Distinction between Learning Curves and Experience Curves and when each model is best – Here

This entry was posted in EV PEV, New Energy Paradigm, Path to a New Paradigm, Storage and tagged , , , , , . Bookmark the permalink.

4 Responses to Experience Curves: How They Work and What They Foretell About Lithium Ion Storage Prices

  1. Pingback: Tesla’s Gigafactory Risk is Exaggerated | The Handleman Post

  2. Pingback: Cost Projections For Lithium Ion Batteries | The Handleman Post

  3. Pingback: Up To Date Cost Curves for Batteries, Solar and Wind | The Handleman Post

  4. Pingback: Why There’s Market Space For 5 Or More Gigafactories By 2020 – Enjeux énergies et environnement

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s