Duke Energy's Poisoned Power
Part II. Insightus Investigates:
How Trustworthy Are Duke's Self-Assessments?
Duke Energy workers inspect damage to one of the company's coal ash dumps at the site of the 2014 spill of 39,000 tons of ash into North Carolina's Dan River (Gerry Broome, AP)
CAMA requires state officials to determine the schedule on which Duke’s ash dumps must be closed, and with what degree of remediation on the utility’s part, based largely on Duke’s own scientific assessments of how much risk each of its 14 dumps poses to public health and safety. Those reports, called 'Comprehensive Site Assessments' (CSAs), which Duke published late last year, are now under review by McCrory’s Department of Environmental Quality (DEQ).
Duke’s CSAs are massive works, each running thousands of pages. Even state regulators find them a hard slog to wade through. As DEQ’s assistant secretary recently warned lawmakers: “Our resources, our staff, are stretched to the absolute breaking point” by the agency’s effort to review Duke’s CSAs. It’s likely that very few people outside of DEQ have ever taken a hard look inside those reports.
But much is at stake here: an estimated ten billion dollars in unwelcome costs if Duke is required to perform the highest possible level of remediation (removal of its ash to lined, dry, closely monitored landfills) at the earliest possible date - or billions in savings if regulators permit Duke to perform much less rigorous cleanup.
With those stakes in mind, insightus data scientists have just completed an in-depth review of one representative example of Duke’s fourteen CSAs, asking whether the company collected and analyzed its data in a scientifically responsible manner, and supported its conclusions rigorously...or whether, instead, Duke may have taken advantage of holes large enough to drive a coal truck through that were built into the CAMA process by the law's Duke-friendly authors.
Our investigation uncovered evidence, presented below, of classic scientific errors like cherry-picking (ignoring data which tends to contradict a desired conclusion), downplaying inconvenient conclusions, assertions unsupported by data, and questionable data analysis methods, all of which — perhaps not surprisingly — have the net effect of de-emphasizing Duke Energy’s responsibility for toxic groundwater contamination surrounding its coal ash dumps.
Allen Steam Station
Duke’s Allen Steam Station (Figure 2), perched just across the Catawba River from North Carolina’s largest city, Charlotte, is the resting place of some 13 to 16 million tons of coal ash in mostly unlined dumps, containing (among many other toxic chemicals) roughly 1.5 million pounds of arsenic (three and a half billion lethal doses), an additional 400,000 pounds of lead, and some 7,000 pounds of mercury (based on EPA assays of similar coal ash).
Figure 2: Coal ash impoundments and neighboring homes at Duke Energy’s Allen Steam Station, Gaston County, NC. Blue outline marks the borders of Allen Station’s ash dumps, including the abandoned (“inactive”; northern half) and still-active (southern) ash basins. Red dashed line marks Duke's property line. To the west, colored circles mark private water supply wells serving homes in the South Point community. To the east, the Catawba River separates Allen’s ash dumps from the city of Charlotte. Blue arrows indicate the direction of groundwater flow assumed by the Allen CSA's authors. Adapted from Fig. ES-1 of Duke Energy’s Allen Steam Station Comprehensive Site Assessment, dated 8/23/2015.
Determining which way groundwater flows at Allen is central to judging the health risk these chemicals may pose to Duke’s neighbors, since flow toward the west from the ash dumps would deliver contaminated water straight to residential wells in the nearby South Point community. Based on little more than the vague generalization that “topography generally slopes from a west to east direction” at Allen, the CSA’s authors propose a hypothetical model in which, conveniently, groundwater flows only away from the site’s neighbors (blue arrows in Fig. 2). But our review suggests another, much more worrisome, possibility, which Duke seems to go out of its way to avoid.
Gauging the Flow
Just as a river flows 'downhill', so too does water moving underground. But the overlying land’s surface contours - those which a stream or river would follow - can bear little relationship to the forces determining which way is ‘downhill’ for underground water. Instead, groundwater flows from a higher water table elevation to a lower one, independent of the overlying land’s surface contours. As explained to us by hydrogeologists we consulted for this investigation, surface water bodies like Duke’s coal ash ponds can locally raise the water table beneath them by acting as a source of added groundwater, while distant wells can lower the water table around them by serving as a ‘sink,’ removing water from the ground - thus potentially reversing the flow of groundwater from what one might expect based on surface topography alone.
To track the invisible flow of water deep underground, hydrogeologists measure the elevation (typically expressed in feet above sea level) of the water table’s surface across a network of observation wells, such as the 98 wells Duke has installed over the years across Allen Station (black circles in Figure 3).
Figure 3: Duke Energy’s observation wells (black circles), and neighboring residential water supply wells (yellow circles) around Allen Station. Brown shapes represent the boundaries of the northern and southern ash basins. Duke's observation wells are labeled with their well IDs, in which ‘S’ indicates a shallow well, ‘D’ a deep well, and ‘BR’ a bedrock well. Closely-spaced wells sharing a single ‘nest’ (for example, AB-34S and AB-34D in the center of the northern ash basin) appear as a single circle at this map’s scale. Observation well geolocations from Allen Steam Station Comprehensive Site Assessment. Residential well locations from NC Dept. of Environmental Quality.
The CSA's conclusion that groundwater flows only eastward at Allen is based on hydraulic gradient calculations (measurements of the water table’s uphill-to-downhill slope, as determined by pairwise comparisons of water table elevations at selected pairs of observation wells). The CSA presents these data in tables like that reproduced, in part, below. But hidden in plain sight in such boring columns of numbers are important insights made much more clear when the same information is displayed graphically, as we have done just below that table, in Figure 4.Here's the way Duke presents the data...
...and here's the right way to look at it:
Figure 4: Water table slopes (hydraulic gradients) from pair-wise comparisons between shallow observation wells at Allen Station. Green arrows represent the pair-wise comparisons Duke's CSA reports in its Table 6-9 (reproduced above this Figure). Each arrowhead points from the higher water table elevation to the lower one ('downhill' for groundwater). Red arrows represent pair-wise comparisons the CSA's authors failed to report (calculated by us using Duke's numeric data), which reveal the water table to the west of Allen's ash ponds sloping downhill toward the neighboring South Point community. Water elevation data from Tables 10-5 and 10-6, Allen Comprehensive Site Assessment.
The green arrows in Fig. 4 indicate all of the shallow well comparisons which the CSA's authors report in their Table 6-9 – all of which indicate water table slopes pointing conveniently (for Duke) away from Allen Station's neighbors. But why use just 9 wells' data, as the authors did, when a total of 45 shallow wells' worth of data were available (black circles in Fig. 4)? As part of our investigation we performed those additional pair-wise comparisons which Duke 'forgot.' The red arrows in Fig. 4 illustrate our results, and provide a possible explanation for why Duke overlooked them – because those additional comparisons lead to an embarrassing conclusion: that groundwater flows in several different directions away from Duke's ash ponds, including toward the west, and thus toward South Point's residential water wells. Figure 4 illustrates how the CSA's authors have engaged in one of the greatest sins of bad science: cherry-picking (selecting only the data that supports a desired conclusion, while ignoring the data that does not).
This more complete picture which our analysis, freed of cherry-picking bias, provides directly contradicts Duke's claim in its Allen CSA that “groundwater...flows...away from the direction of the nearest public or private water supply wells” at Allen. And that, in turn, contradicts its headline conclusion that “no imminent hazard to human health” is posed by Allen Station's ash dumps.
Like “the dog that didn’t bark” in the Sherlock Holmes mystery, Silver Blaze, what’s most important — and most damning — in Duke’s Allen Station report is what its authors choose to remain silent about: the very real likelihood that contaminated groundwater flows from Allen Station’s ash dumps toward neighboring residential wells, as revealed by Duke’s own data.
Q: When Is a Poisoned Well Not a Poisoned Well? A: When a Polluter Has Friends in High Places
CAMA's requirement for Duke to perform these self-assessments is accompanied in the law by its parallel requirement for the state's Department of Environmental Quality (DEQ) to test residential water wells located within half a mile of Duke's dumps for contamination with coal ash-derived chemicals. DEQ’s test results for more than 400 such wells are available on its web site (although not in an easily digestible form).
Figure 5, below, charts the results of DEQ’s residential well tests around Allen Station for the toxic coal ash contaminants chromium-6 (Cr-VI, a known carcinogen; top panel in the Figure) and vanadium (V, a suspected carcinogen which also causes anemia and neurological effects; bottom panel). As elsewhere throughout this report, here we present the original data in a graphical form making its significance easier to grasp.
Figure 5: Average concentrations of chromium-6 (top panel) and vanadium (bottom panel) in residential wells neighboring Allen Station. Concentrations are expressed in micrograms per liter (ug/L), and are color-coded as shown in each panel's legend. Several wells shown here were sampled repeatedly on different dates by DEQ; each concentration shown here is the average of all sampling events. The minority of wells which consistently tested negative for these chemicals are not shown here.
As Figure 5 shows, residential wells in the South Point neighborhood west of Duke’s ash dumps are widely contaminated with these two toxic chemicals. As would be expected if Duke’s coal ash is the source of this contamination, the most heavily poisoned wells (red through light green circles in Fig. 5) are generally closer to Duke’s ash basins, while more distant wells are generally less heavily contaminated (dark green through blue circles).
Of the 476 DEQ-tested residential wells neighboring Duke’s fourteen ash dumps across the state, Department of Health and Human Services (DHHS) epidemiologists found the test results sufficiently worrisome to move them to issue ‘do-not-drink’ advisories to 424 of their owners, most often due to chromium-6 and vanadium levels exceeding the department’s guidelines. Understandably, those notices struck fear in the hearts of many recipients. In the words of one young mother, Sherry Gobble:
I just remember shock, feeling shock, and I couldn't hardly hear what they were saying to me because I was replaying in my mind every time I had fixed a jug of Kool Aid, every time I had made a pot of tea. So I couldn't even pay attention to what they were saying because, as a mom, all I could think about was, "What have I done? What have I allowed my children to drink?" And just sheer fear went all over me.
On June 1st of last year, just as such test results began to come in, Duke Energy’s CEO, its general counsel, and the president of its North Carolina business unit met secretly for dinner at the governor’s mansion with Duke’s ex-employee Gov. McCrory, along with his secretary of the Dept. of Environmental Quality and the state’s own general counsel. Duke and McCrory have both declined to reveal what they discussed at that unprecedented secret dinner, except for the state press secretary’s bland assurance that it was “a constructive dialogue with Duke Energy officials [including] topics about the economy, the environment, energy and job creation.”
While we are not permitted to know the details of Duke’s and McCrory’s dinner conversation, we may see hints in more recent events. On January 13th of this year, DEQ’s Assistant Secretary for Environment, Tom Reeder, complained to the state legislature’s Environmental Review Commission that DHHS’s chromium-6 and vanadium standards - which triggered those warning letters - were in his opinion overly cautious; so much so, Reeder claimed (without providing evidence), that "over 70% of public water systems in the United States" would likewise exceed those standards.
Perhaps Reeder’s concern was a matter of optics: maps of North Carolina (like those of Fig. 5) highlighting Duke’s coal ash dumps ringed with residential wells serving up water the state deems undrinkable are bad for business, and thus bad for Reeder’s boss, the governor. Or perhaps the governor’s friends at Duke were concerned that such findings might expose the corporation to lawsuits by the affected homeowners. But to be fair, it’s a complicated story. For, as Robin Smith (former Assistant Secretary of Environment under McCrory’s predecessor, Democratic Gov. Bev Perdue), explained to us in an interview:
There is no existing state groundwater standard for vanadium or chromium-6. If there isn't a drinking water standard, an interim standard - called an IMAC - may be calculated based on review of human health effects such as toxicity or increased cancer risk. In North Carolina, IMACs are usually developed by epidemiologists in the Department of Health and Human Services’ Division of Public Health.
DEQ asked the Division of Public Health to develop IMACs for both vanadium and chromium-6, to be used in assessing wells around the coal ash ponds. Division of Public Health calculated an IMAC for each that was based on 'incremental increase' in cancer risk -- the concentration that would cause 1 additional case of disease per 1 million individuals, which is the threshold generally used in both state and federal programs.
I think the practical problem in this case is the disconnect between the resulting IMACs and state/federal drinking water standards - which meant that a well owner could receive a "do not drink" order even though the water would meet Safe Drinking Water Act standards applied to public water systems.
Just this month, McCrory’s DEQ finally prevailed over DHHS, persuading health officials to withdraw nearly all of their do-not-drink advisories for private wells near Duke ash dumps. Well owners like Allen’s neighbor, Amy Brown, were left understandably confused:
How can my water be unsafe yesterday but today, with the standards still the same, you want to tell me my water is now safe? I don’t feel that anyone has given me any information for me to not fear my water.
DEQ did not respond to our request for data substantiating Reeder's claim that "over 70% of public water systems in the United States" would not meet DHHS's interim standards for vanadium or chromium-6. But we have done the next best thing, comparing DEQ’s measured chromium-6 and vanadium levels for South Point’s residential wells with data reported recently to EPA by the municipal water systems of four of North Carolina’s largest municipalities. Table I, below, compares those cities’ 2013-2014 chromium-6 and vanadium averages with those for Allen Station’s residential neighbors.
Table I: Comparison of average vanadium and chromium-6 concentrations for four North Carolina municipal water systems (2013-2014) with average Allen neighborhood residential well water concentrations. Allen data are from DEQ’s well water testing web site. Municipal systems’ data are found at these URLs: http://charmeck.org/city/charlotte/Utilities/PublicationsandEducation/Pages/waterqualityreports.aspx, http://www.greensboro-nc.gov/modules/showdocument.aspx?documentid=27867, http://www.greensboro-nc.gov/modules/showdocument.aspx?documentid=24509, https://www.raleighnc.gov/content/PubUtilAdmin/Documents/CCR.pdf, http://gsa.raleighnc.gov/smb/ptlprdapp1/PTLPRD/PubUtilAdmin/Documents/CCR/Archive/CCR20131231.pdf, http://www.cityofws.org/waterqualityreport2014.
Our analysis in Table I supports DEQ’s argument that municipal water systems would frequently exceed the interim standards for vanadium and chromium-6 which DHHS developed for the state’s CAMA-mandated coal ash studies (0.3 and 0.07 micrograms per liter, respectively). But Reeder’s argument nevertheless conceals an important point which Table I makes clear: whatever standard the state may choose to employ, Duke’s neighbors around its Allen Steam Station are now being encouraged by McCrory’s DEQ to drink well water contaminated with chromium-6 and vanadium at levels which exceed nearby cities’ drinking water by, on average, 20- to 30-fold. And in the case of the few most heavily contaminated wells near Allen Station, that difference is an even more alarming 60- to 140-fold (not shown in Table I). In its announcement lifting the state's do-not-drink advisories, McCrory's DEQ misleadingly assured Duke's neighbors that their well water "is as safe to drink as most cities and towns across the state," when in fact, as Table I demonstrates, their well water presents them with risks of developing cancer which may be tens to hundreds of times higher than nearby city-dwellers face.
There's a common name for the charade of setting limits, then changing or even abandoning them altogether if they prove to be inconvenient: moving the goalposts. And it's yet another classic trick of bad science which Duke and McCrory's overtly politicized DEQ have pulled from their sleeves in their quest to short-circuit North Carolina's coal ash management law.
Finding 'Clean' Water: The Question of Backgrounds
Our investigation has also uncovered examples throughout Duke’s site assessments of important assertions of fact made without any scientific support. Perhaps the most significant example we encountered concerns the critical question of how to measure ‘background’ water quality - the chemical composition of groundwater which has not been contaminated by Duke’s ash pits.
Background water quality is an important concern because heavy metals like vanadium and chromium can occur naturally in some rocks and soil types, so these chemicals can be found - at least at very low levels – in even unpolluted groundwater. Indeed, Duke has insisted repeatedly that high levels of chromium-6, vanadium, and other toxic groundwater contaminants in and around its ash dumps might be naturally occurring chemicals from the soil itself, rather than products of its ash pits – in other words, ‘background’ levels of contamination for which Duke cannot not be held responsible.
In the case of the Allen Station site assessment, the CSA’s authors employed newly drilled wells designated BG-1, BG-2 and BG-3 (to the west and south of its ash basins; see Figure 3) as ‘background’ wells - but offered no evidence to support their assertion that those wells sample true background water, rather than ash-contaminated water.
That lack of evidence is concerning in light of Duke’s documented history of choosing background well locations unwisely. In the Allen CSA’s own somewhat disjointed words, referring to its 2010 best guess regarding a previous ‘background’ well, AB-1R (emphasis added):
Existing...well AB-1R has been considered by Duke Energy to represent background water quality at the site since being installed in 2010 [report p. 65]
Sampling data since March 2014 for existing background monitoring well AB-1R and the groundwater flow direction determined from the CSA activities indicate this well may be influenced by the ash basin [report p. 107]
Here, the phrase “influenced by the ash basin” is Duke's delicate but oblique way of saying “contaminated by the ash basin,” amounting to the company's admission that it has in the past guessed spectacularly incorrectly regarding where to locate a background well at Allen. But with no better evidence today, Duke assures us that – this time for sure – its new background wells BG-1, 2, and 3 are reliably distant from the pernicious “influence” of its ash dumps (emphasis added):
The BG [new background] locations were strategically placed to maximize physical separation from the ash basin in areas believed not to be impacted by the ash basin [report p. 65]
The newly installed BG wells are not located hydraulically downgradient of the ash basin, and are likely representative of background groundwater quality conditions at the site [report p. 65]
The word “believed” here is more than a little alarming, because ‘belief’ is not a word in the vocabulary of technology: scientists are tasked with producing data, not beliefs. And not only does Duke offer no data to support its belief, but as we have already shown, Duke’s own data puts the lie to it, since the shallow ‘background’ wells BG-1 and BG-2 may, in fact, be hydraulically downstream of ash basin wells AB-20, AB-37, and AB-38 (as we have already shown in Fig. 4) according to the hydraulic gradient calculations Duke chose not to publish.
This is more than a mere technical quibble. If Duke Energy’s supposedly clean ‘background’ wells are, instead, contaminated with heavy metals from its ash dumps, that creates the appearance that the contamination found by DEQ in nearby residential wells is Mother Nature’s fault, not the corporation's. Determining true background levels for these contaminants is fundamental not only to determining the risks Duke’s dumps pose (and, thus, the degree of remediation which the state should demand from Duke). As well, this issue is also central to the question of Duke’s potential legal liability for contaminating its neighbors’ wells. For as Robin Smith explained to us in our interview with her (emphasis added):
There have not been many well owners who have sued for damages like loss of property value. In that kind of lawsuit, the plaintiff has to prove the defendant caused the damage, which means ruling out naturally occurring contamination and showing a link between the defendant's activity and the contamination affecting the well.
Badly chosen background wells are in Duke Energy’s best interest, so its decisions regarding where to site such wells should be backed up with solid data – data the company does not provide in its site assessments.
Garbage In, Garbage Out: When Good Models Go Bad
In the end, state regulators’ final decisions regarding the fate of Duke’s ash dumps will be based largely on detailed computer models of the predicted future spread of groundwater pollution from those sites – models which CAMA requires the company to provide to the state as part of its site assessments.
Computer modeling is a powerful means of understanding large, complex data sets and employing them to predict future events, used successfully today in everything from weather forecasting to aircraft design. But the first rule of the modeler’s art is “garbage in, garbage out:” a model is only as good as the assumptions, design, and data that go into it. We have already seen that some important components of Duke’s data, such as its hydraulic gradient calculations and its background measurements (all of which feed into Duke’s computer models) might, in fact, be garbage (in the modeler’s sense of that word). But our investigation has also uncovered evidence that even certain basic design features of Duke’s models themselves appear to be simply wrong.
Sources and sinks: When modeling groundwater flow, getting the so-called ‘sources and sinks’ right (the locations where water enters or is removed from the ground) is essential.
According to the Allen computer model’s own authors (writing in Appendix B of CAP-Part 2 of the Allen CSA), Duke’s groundwater flow model includes just 44 private water supply wells within the model’s bounds (each constituting a ‘sink’ in the model). However, our geospatial analysis of DEQ’s database of private wells surrounding Allen Station identified more than twice that number of private wells – a total of 93 – within the model’s bounds.
Including an unrealistically low number of sinks within the at-risk South Point neighborhood in its groundwater flow model — as Duke appears to have done — stacks the deck in the company’s favor, because actively pumping wells will lower the water table in their vicinity (thus promoting the flow of contaminated groundwater in their direction). If the model underestimates that effect by including too few such wells, then it is likely to underestimate the flow of ash-contaminated groundwater toward residential wells, and thus to underestimate the risk Allen Station presents to its neighbors’ health and safety.
Independent Review: Another important step in modeling is calibration - ‘fine-tuning’ the model to achieve the best possible performance. Calibration is, unavoidably, a rather subjective process, presenting the risk that the modeler might, either consciously or unconsciously, bias the model’s performance by the choices he makes during the calibration process. In some applications, such as weather forecasting, that risk is mitigated by the modeler’s powerful motivation to get the model right (i.e., to produce predictions which will prove accurate). But in other applications, where the modeler may be influenced by other motives...where, for instance, he might have a vested financial interest in squeezing ‘good news’ rather than ‘truth’ from his model...the modeling process benefits greatly by having disinterested outside experts independently review and approve the model. Duke’s computer model for Allen Station, as well as its other North Carolina ash dumps, was independently reviewed by the non-profit Electric Power Research Institute (EPRI).
But who reviews the reviewers? Is EPRI the right source for independent reviews double-checking Duke’s work?
EPRI, an industry-sponsored organization, derives the lion’s share of its nearly $400 million in annual revenues from dues paid by its electrical utility members, including Duke Energy, America’s largest privately owned electrical utility. Member-companies’ dues are calculated on a sliding scale based on each member’s total assets, so Duke is likely one of EPRI’s highest-paying members — a status which also earns it a seat on the institute’s board of directors. In effect, Duke Energy is both one of EPRI’s largest customers and one of its bosses.
EPRI’s qualification to serve as an objective reviewer – a reasonable question given its financial interest in Duke’s success – is brought further into question in light of our investigation’s finding of deficiencies in Duke’s Allen Station groundwater model which EPRI’s review fails to call attention to.
Among the numerous technical issues which EPRI’s scientists considered in reviewing the Allen model (in Appendix B of CAP-Part 2 of the Allen CSA), few are more important than the models’ residuals – modeler-speak for the differences between a model’s predictions and the observed real-world facts. For example, if a weather-forecasting model predicts that today’s high temperature will be 90 degrees, when the actual high proves to be 70 degrees, the model’s residual is 20 degrees. Residuals are an inescapable fact of life because no model is perfect, but in a properly calibrated model they should be reasonably small and randomly distributed across the model's bounds if a model is to perform successfully.
Figure 6 is our geospatial analysis of the Allen model’s groundwater elevation residuals across the model's space - the differences between predicted and actual groundwater elevations as reported by the model's authors. Each vertical bar in Fig. 6 represents the magnitude of the residual at the shallow monitoring well location indicated by the dark gray oval: green upward-pointing bars indicate predicted groundwater elevations which are higher than true elevations, while red downward-pointing bars indicate predicted values which are lower than actual measured values.
Figure 6: Spatial distribution of shallow well residuals in the Allen groundwater flow model. In this figure the map is rotated vertically to present a nearly edge-on view. The gray line indicates the spatial boundaries of the Allen computer model. Gray ovals indicate the locations of all shallow observation wells at Allen Station. Green bars pointing up from these locations indicate positive residuals (predicted water levels higher than actual levels), while red bars pointing down from their well locations indicate negative residuals (predicted water levels lower than actual levels). Gray ovals without accompanying green or red bars indicate well locations for which no residuals were reported. Predicted and actual water elevations used in our analysis are from Table 3 of CAP-Part 2 Appendix B of the Allen CSA.
This way of looking at the data reveals two serious flaws in Duke’s computer model for Allen Station:
1. Residual errors are not randomly distributed. The model consistently overestimates groundwater elevations on the western, eastern, and northern boundaries of its domain (most notably in and around the important South Point residential neighborhood), while underestimating groundwater elevations at the center of the ash dumps. In effect, the model believes (incorrectly) that the coal ash basins sit at the bottom of a bowl, while sensitive sites like the South Point neighborhood sit higher up on the rim of the bowl. The effect of this on the model’s predictions will be to cause it to underestimate the rate of contaminant migration from the ash basins toward Allen’s unlucky South Point neighbors, and thus to suggest that Allen presents less risk to those neighbors than it actually does.
2. The model's authors present residuals for only a fraction of the site's wells. Each gray oval in Fig. 6 marks a shallow observation well with available water elevation data reported in Duke’s site assessment. Ovals without accompanying red or green residual bars are wells for which residuals were not reported, suggesting that those wells' data was not included in Duke’s model - defying the fundamental rule of modeling that the more real-world data a model includes, the better. Duke’s modelers appear to have chosen to ignore data from three of the five wells in and around the threatened South Point neighborhood, as well as five of the nine wells within the southern (active) ash basin, the major source of pollutants at Allen Station.
EPRI’s review of the Allen Station model fails to point out either of these two critical flaws. Given that both of these omissions work in Duke’s favor, it is tempting to suspect that the reviewers' silence on these points is of the 'if-you-can’t-say-something-nice' variety. It's worth noting that EPRI’s reviews of groundwater models for some other Duke sites, such as that for Buck Station, approvingly note that “There is no significant spatial bias in the distribution of residuals.” But in their review of the Allen Station model, where obvious spatial bias is present, the reviewers are simply silent.
EPRI scientist Bruce Hensel, the Allen model review's lead author, did not respond to repeated invitations to comment for this article.