15 June 2012

Economies of scale

Log-Log Fig 3.5 from EPA (1980) shows IRS...
JD sent me some interesting reports on economies of scale in wastewater treatment plants.*

As most of you know, "increasing returns to scale" (IRS) refers to the increase in efficiency (or decrease in cost per unit of material handled) that you get from a larger facility. Greater efficiency must, however, be balanced against decreasing scales from, e.g., moving material over greater distances to a larger facility, less local customization, etc.

The water business has been characterized by IRS for many years (due to the low cost of energy, ease of digging up streets, importance of discharging water at a single point, etc.), but IRS makes less sense if treated wastewater is going to be used again (IPR, or water recycling), energy is expensive, centralized management is less responsive, etc.

Bottom Line: Make sure you include ALL costs before you go for an "efficient" scale facility.

* Check out these reports if you want more details:

EPA (1976) An analysis of construction cost treatments for wastewater treatment plants [PDF]
EPA (1979) Determining wastewater treatment costs for your community [PDF]
EPA (1980) Construction costs for wastewater treatment plants 1973-1978 [PDF]
MT DEQ (2007) Wastewater treatment performance and cost data to support an affordability analysis for water quality standards [PDF]

4 comments:

The Engineer - Dr. Steven D. Sanders said...

Great blog and your book is very good. Blog post regarding your book

http://witandwisdomofanengineer.blogspot.com/

David Zetland said...

Thanks Steven!

Eric said...

'Including all the costs' is now my mantra.

As to the plot, the statistician in me is disgusted. A least squares fit to log/log data is a bad idea. Not including the error bars associated with each point makes it worse. You should fit the original data with an appropriate function and include the errors from each point. Then you can convert to a log-log scale for display. As to the display listed in the post, Edward Tufte would suggest that the person who made the graph should read Tufte's books about communicating data effectively.

David Zetland said...

@Eric -- I agree. That plot violates everything in Tufte, but it WAS made in the 1970s. Progress, right? :)