Lawrence Berkeley National Laboratory Responds To Criticism Regarding Recent Roofing Study

This article is written in response to the article, Analysis Of Berkeley Labs Study Details Flawed Science, posted on February 25, 2014.

By Benjamin Mandel

As an author of a recent study that compares the economics of white, green, and black flat roofs in the United States, I am pleased to have this opportunity to respond to critiques of our work. Many of the objections raised in a recent review of our paper by the EPDM Roofing Association (ERA) are valuable insights from experience in the roofing profession. However, some of its critiques betray a misunderstanding of the scope of our paper, as well as the methods we used to reach our conclusions.

We set out to provide guidance on a common question–what kind of roof should I install?–in very broad strokes. To do this, we assessed the economics of three flat-roof strategies–white roofs, green (or vegetated) roofs, and black roofs–using data reported in 22 case studies in which a white or green roof was considered against a black roof. The types of roofs within each color category varied between cases, as did individual project circumstances. Because of this natural variability, we relied on median values of the data within each color grouping. This strategy allowed us to produce what we think of as comparisons between generic white, green, and black roofs.

But it would be irresponsible to prescribe one single figure based on generic roofs as the net savings of a roof color comparison on a nationwide scale. In light of the sizable variability in climate, economy, and policy throughout the country, our paper repeatedly emphasized that this generic comparison should be supplemented with project-specific information. For instance, the ERA review panel asserts that we should have factored ballasted EPDM systems into our analysis; I agree that these systems can make sense in certain conditions (see here for a sound overview of this topic), but no roofs in our case studies used this approach. If a particular roofing project is considering using a ballasted roof, the scales may tilt depending on local conditions.

Similarly, the panel lamented the assumed 20-year lifetime for black roofs because EPDM has been shown to last more than 30 years, while disputing that white roofs last as long as our chosen 20 years. In fact, these service life assumptions were taken from our case study data–as with the rest of our data, we used median values of the service live assumptions that were reported for each project. (For what it’s worth, Facilities Management at UC Davis finds that their white PVC roofs outlast traditional roofs–one such membrane was installed more than 30 years ago and is still in service today.) Still, if a building owner knows the EPDM membrane he/she is thinking of using will last 35 years rather than 20, this is very relevant information to include in his/her decision. This is what we mean when we say that this decision needs to be evaluated on a project-by-project basis.

ERA’s review attempts to discredit our findings by claiming that our methods violate the scientific method, producing results based on “flawed” or “faulty” science. But I will freely concede that our study contributes no new science to the literature–we conducted no experiments, nor did we claim to. Our study contributes to the scientific literature by synthesizing established research and cost-benefit analyses from actual projects to construct a holistic life-cycle cost analysis that accounts for such factors as the direct global cooling effect of roof reflectance. So the assertion that we have violated the scientific method seems misguided–there is no experiment in this study. We were just doing what we could with a small sample of illustrative data.

With that said, we agree with ERA’s assessment that the sample size in our study is smaller than would be ideal. While we would have preferred thousands of data points, our considerable efforts could only muster 22. Even within these few examples, however, we observed fairly wide variation. We attempted to handle this variability sensibly, by using median values from our sample data, as medians are invariant to outliers. This is important to note in light of accusations that we have stacked the deck against black roofs by including cases with very high costs for black roofs, for instance. By using medians rather than averages, the data from less representative projects (on both high and low ends) do not factor into the ultimate economic results, though they do provide valuable insights for real-world roofing project considerations.

An unfortunate downside of our case study approach is that we were dependent upon the quality and completeness of the case studies we drew from. Regrettably, none of these case studies reported the added energy costs associated with heating buildings beneath white roofs. This “winter heating penalty” is an established phenomenon and our intent was not to wish it away–it is important to consider ways to conserve the most energy over the whole calendar, not just during summer.

As far as black roofs are concerned, our generic results show white roofs to be more cost-effective over 50 years, if only by a bit. Still, in some of our cases it appeared that a black roof would have made more economic sense than a white membrane. This would be just fine if economics were the only relevant input into the decision-making process. However, roofing choices can also influence some factors that are hard to quantify and capture in an economic analysis, such as the urban heat island effect.

Importantly, we acknowledge the possibility that in a Northern city like Chicago, which has cold winters but is susceptible to dangerous summer heat waves, the economics alone may favor a black roof. But because the dark color of black roofs leads them to absorb up to 95% of incoming sunlight and use that energy to heat cities, their use may exacerbate heat waves and therefore pose a public health threat (see this recent article for more on this point).


In light of this potential misalignment between financial considerations and public health goals, we suggest that policy should intervene to require cool roofs in these cases. This may not be a popular solution to everyone, but to the extent that cool roofing can keep cities more comfortable during a heat wave, that’s a benefit worth capturing.

Mandel is a Research Assistant in the Heat Island Group at Lawrence Berkeley National Laboratory, which studies the science of heat island mitigation strategies including cool roofs, cool pavements, and urban vegetation. For more information on the Heat Island Group, visit


  1. It is hard to evaluate the merits of the arguments, because the original study was published in a journal that doesn’t provide open access. All research from a public institution such as the Lawrence Berkeley National Laboratory should be published in an open access journal. After all, if the American public pay for the LBNL, they should be able to read its research for free.

  2. This may not be a popular solution to everyone, but to the extent that cool roofing can keep cities more comfortable during a heat wave, that’s a benefit worth capturing. If a particular roofing project is considering using a ballasted roof, the scales may tilt depending on local conditions.

Comments are closed.