Completion of the 2080MW Hoover Dam on the Colorado River in 1935 was a significant moment in the history of dam building in the US. Some say that it was a tipping point that fired the imagination of dam builders and led to a surge of dam building across America [1]. However smaller scale hydroelectric dams were being built all over the US from the turn of the century. What building the Hoover Dam really fired was the confidence of dam-builders: if the Colorado River could be tamed, then they could dam any river.

However not everyone considered the dams to be progress, and over the decades movements to prevent new dams being built, and to remove old ones, have grown stronger.

Recently, works have commenced to prepare for removal of the Elwha and Glines Canyon dams in Washington State. These dams are relatively tall: the Glines Canyon dam is 64m high, but their power output is small, with a combined total of 28MW. What really sets them apart however is the cost of their removal. A price tag of over US$300M sets a new high-water mark in the price that the nation is prepared to pay to restore it’s rivers.

So does this represent a tipping point in the opposite direction – towards removal of dams?

Another example, this time from California, might suggest that a tipping point has been reached. In November 2008 an agreement was signed between owners and protagonists concerning the eventual removal of four dams on the Klamath River. Sceptics point out that the Klamath River Agreement is really an accord to keep the dams until 2020, to study the feasibility of removal, and for the state and utility to split the costs if the dams are eventually removed [2]. However, even sceptics would admit that the agreement sets the direction [3]. The eventual removal of the Klamath dams would represent an important victory for dam removal activists, especially in a state with ambitious goals in renewable energy generation.

The US has approximately 79,000 significant dams [4] and probably has the most experience in removing them. Over the twentieth century 467 dams were reported to have been completely or partially removed across the US [5] and the total number of removals now stands at over 600 [6]. The pace of removal is reported to have picked up over the decades, coinciding with the 50 year licensing cycle under Federal Energy Regulatory Commission (FERC) rules [7] (Table 1 and Figure 1).

The average size of removed dams appears to have decreased in recent decades. This may be due to an increase in funding for removal of small dams over this period. For example American Rivers administers a grant programme for the removal of small dams as part of the National Oceanic and Atmospheric Administration (NOAA) Open Rivers Initiative.

In an analysis of 417 case studies Pohl [8] found that environmental reasons were most commonly cited for dam removal (39%), followed closely by safety (34%).

Hydro dam removals

There is no single source of useful and accurate information on hydroelectric dams removed in the US. The information has been collated from a range of official sources – such as the FERC, state registers, and the US Army Corps of Engineers and from unofficial sources, such as the Stanford University National Performance of Dams Programme and University of California Berkeley Clearing House for Dam Removal Information, Idaho National Laboratory and from dam removal activist literature (such as American Rivers). Where possible the information has been cross checked against original sources.

Our research found reasonably complete information for 17 of the 23 hydro dams reported as having been removed to date. A further two removals are underway (counting the two Elwha dams as a single scheme) and six were classified as “planned”, including the four Klamath dams. Dams were only entered into the “planned” category if the dam owner has agreed to remove the dam, or been ordered to remove the dam, within a set timeframe. Counting dam removals planned and underway, the data set consists of 25 reasonably complete sets of data (Table 2) [9].

Several observations are immediately apparent from the data on hydro dams removed to date:

• The dams tend to be of moderate height (range 5m to 18m).

• Most of the dams had a small installed generating capacity (0.4 to 10MW).

• The dams are reasonably old (average age 87 years at removal).

• Many of the hydro schemes had already been retired at the time of removal (86%)

While the removed dams are often small in generating capacity, comparison against data for hydro dams recognised by FERC reveals that that, the size distribution of removed dams is reasonably representative of the stock of dams in the US, at least for dams below 100MW installed capacity (Figure 2).

The decommissioned hydro dams represent 1.8% of the total number of FERC recognised hydro schemes, but the total generation capacity of 40MW decommissioned to date represents only 0.05 % of the 78,000 W total installed generating capacity of hydro schemes in the US [10]. This is because a small proportion of large dams provide most of the installed capacity and none of those have been removed. Dam removal projects currently underway will increase the total generating capacity removed to over 100MW.

Whether a tipping point has been reached is a moot point, but the data is consistent with an emerging trend towards: removing larger hydro schemes with greater installed capacity; removing operating hydro schemes, not just retired hydro dams

Sediment removal

Sediment management is a major consideration in dam removal, and probably the most significant issue on most decommissioning projects. In recent times there has been a trend to use more natural processes to manage sediments (Box 1). Current examples include the Condit Dam and the Bull Run hydro scheme removals, both of which are due to be completed in 2009 [11].

Cost of dam removal

Historical costs of dam removal were inflated to 2008 US dollars using the US Army Corps of Engineers construction cost index for dams [17]. To allow a range of project sizes to be compared, removal costs were also expressed as a percentage of the construction cost to build an equivalent facility (Table 3). A hydroelectric project cost model developed by the US Department of Energy [18] was used to estimate current US construction costs for those hydroelectric projects. The model is based on only one input variable, being the installed electricity generation capacity, so the estimated costs are not site-specific and are indicative only. Nevertheless the data suggests that removal costs are typically 5 to 50% of construction costs.

There is some evidence that hydro dam removals are getting costlier. For instance prior to 1999, removal costs were typically less than 10% of the cost of building an equivalent hydroelectric scheme of the same installed generation capacity. Since 1999 the cost of dam removals has increased, typically costing 20-40% of new construction costs. It is possible that the early removals were biased towards the easiest dams to remove, where low costs made it relatively easy to justify a removal decision. While local factors will affect any removal decision, it is logical to expect that dams that are inexpensive to remove will tend to be removed earlier than others.

One cost stands out in Table 2, being the Elwha dams removal which is being conducted by the US National Parks Service. This is clearly an outlier when compared to the other data. Reasons why this removal project is so expensive are examined in Box 2, but can be summed up in a single word: sediment.

Factors affecting cost of removal

Regression analysis was used to indicate the most significant factors influencing cost of dam removal. Elwha data was excluded because it is an outlier, and the four Klamath dams were excluded because of the uncertainty regarding the timing of their eventual removal and due to the wide variations in projected removal costs reported in the literature. All costs for the 18 remaining dams in the data set were brought to 2008 dollars using the USACE cost index.

Simple linear regressions indicated that installed capacity, and dam height were the most significant factors. A multiple linear regression analysis found that these two factors combined accounted for 83% of the variation in removal cost.

Estimates of reservoir area were only available for nine dams. A simple regression analysis tentatively indicated that reservoir area is important, ranking below installed capacity as a predictor of cost, but ranking above dam height.

Factors influencing dam removal

It has already been seen that most hydro dams removed to date had already been retired. Clearly this is a good predictor of likelihood of removal. Other factors influencing the likelihood of dam removal were expected to include:

• Size – smaller dams being may be more likely to be removed.

• Rank – smaller dams in a portfolio owned by an operator could be more likely to be removed (for example as a form of mitigation for continued operation of others).

• Region – dams on pacific salmon rivers may be more likely to be removed.

In fact, when compared to the FERC database, no significant overall difference was found in likelihood of a dam being removed due to the size (Figure 2), at least for dams below 100MW installed generating capacity. Rank was also examined but only a weak relationship was found, which may have simply been chance. The overall historical data showed no compelling evidence of regional differences in the likelihood of dam removal (Table 4). However closer inspection of the data shows that many of the early hydro dam removals were on the Atlantic coast, but most recent and planned dam removals are on the Pacific coast. This emerging trend may indicate that activism focused on Pacific salmon rivers is beginning to have an effect. In summary none of these additional factors were found to be good predictors of the likelihood of a dam being removed.

Timing of removal of operating dams

When the case histories are examined in depth, as part of a process of relicensing, then an interesting pattern emerges. What appears to happen, time and again, is that the owner agrees to decommission an operating dam and, in return, the dam is allowed to run on for a number of years to pay for it’s own removal.

This achieves a classic win-win: the river restoration activists win by getting another section of the river restored, and the portfolio owner wins because, having accepted that the dam will be removed, the hydro then generates it’s own removal fund. Savvy hydro portfolio owners use the removal of one dam to burnish their environmental credentials, while ensuring that the vast majority of their hydro generation capacity portfolio is relicensed.

Summary of hydro removals

While a large number of dams have been removed in the US, this has started from a very large base, in a country with a long history of industrialisation.

The arguments for planning the removal of dams are cogently summarised by John Seebach of American Rivers: “Dams are tools, but like all tools, they eventually wear out and stop serving a useful purpose: even a revenue-generating benefit like hydro power doesn’t always outweigh the cost associated with a dam’s environmental impacts or public safety hazards. When these costs begin to outweigh the benefits, it’s time to take a serious look at decommissioning. Removal should always be an option in relicensing, but since many of our hydro dams are still quite functional and produce benefits that are deemed worth the costs, it’s not going to be a serious option in the majority of hydro licensing cases.

“The problem with hydro dams is that, even though they’re designed with a non-permanent 100-200 year life span at best, they aren’t planned that way: the question of what to do with them when they’ve outlived their usefulness is one that’s not dealt with seriously during the initial permitting and construction of a dam, which is assumed to be a permanent fixture. As a result, when a dam needs to be removed, the taxpayer is often stuck with the bill, even if the dam was constructed, owned, and operated by private investors who profited from that investment. We need to consider the entire lifecycle of infrastructure when we decide to build it, rather than pushing that cost onto future generations of taxpayers.”

Removal of the Elwha dams and the recent Klamath Dams Agreement may represent a tipping point, where the appetite for dam removals will rapidly grow. However it is more likely that it represents a rebalancing of the relative weight given to electricity generation, recreational and environmental considerations.

Kevin Oldham, Director, SPX Consultants Limited, PO Box 25 953, 11A Polygon Rd, St Heliers, Auckland, New Zealand. Tel: +64 9 575 5758. Email:

1. Removal of two dams in Michigan

Sturgeon Dam

The 16m high Sturgeon concrete arch dam was built on the Sturgeon River, Michigan, in 1919 to supply 0.8MW of hydroelectric power. In 1998 the dam owner agreed to remove the dam, stating that it was no longer economic to operate [13]. The dam was progressively removed in three stages of 5m, 5m and the remainder, on the first, third and fifth years of the removal programme [14]. Removal commenced in 2003 and was completed in 2007 [15] at a cost of approximately US$2M [16].

Stronach Dam

The 10m high Stronach Dam was built on the Pine River in Michigan in 1912, generating 2MW through an adjacent powerhouse. High sediment loads in the Pine River filled the 27ha reservoir just 18 years later, in 1930. The hydro plant operated for another 23 years, but was retired in 1953 [12].
The Stronach powerhouse was demolished in 1996. A series of stoplogs were inserted and progressively removed: one 150mm high stoplog per quarter, until the river was back down to its natural level by 2003. This gradual removal was aimed at reducing environmental effects downstream of the dam, by gradually releasing stored sediment back into the downstream river. Riprap was placed in some locations within the reservoir to prevent the accumulated sediment from eroding too quickly.

2. Removal of dams on the Elwha River

The 33m high Elwha and 65m high Glines Canyon storage dams produce a combined total of 28MW of electricity, but they also isolate spawning areas in the headwaters of the Elwha River for several threatened fish species. These privately-owned dams were built in 1910 and 1926, predating formation of the Olympic National Park in which they are now located.
The US National Parks Service is proposing to spend a total of US$308M to purchase and remove both dams by 2012, and has already commenced construction of downstream mitigation works, in preparation for dam removal [19]. The first contracts are for improvements to water supplies to downstream communities, which will be adversely affected by high sediment loads in the first few years after dam removal. In fact the cost of removing the dams and works to stabilise silt in the Elwha reservoirs represents a minority of the overall cost (see figure opposite). Excluding purchase of the hydro schemes, the estimated cost of the Elwha removal project is 181% of cost of building equivalent new hydroelectric power plants.
Much of the Elwha project costs are for downstream mitigation works, such as building new water treatment plants to cater for elevated sediment loads and replacing septic tank systems in floodplains, which will now be inundated again. These mitigation works exceed the cost of removing the dams. A major factor is that the proposed sediment management method is to let the river erode the accumulated sediments over a relatively short period of time. Over the first three years following dam removal it is expected that high sediment loads will affect the quality of water supplies for downstream communities and industrial users. Sediment will also temporarily raise the river bed levels, requiring additional flood protection works [20].
This sediment management solution was chosen by the National Parks Service in 1995 because it was cheaper than dredging the reservoirs to remove the sediment. The US National Parks Service has an aggressive policy of removing such dams from national parks [21] and the federal declaration of several species of fish present in the Elwha as threatened, and other factors may have precluded a slower and possibly cheaper, staged removal process.


Table 2
Table 3
Table 4
Table 1