Reply:

Agreed. See added detail. We have added the spreadsheet equations and have detailed the important calculations in the text as well.

Need to consider uncertainties for various input parameters and how those uncertainties affect conclusions.

Reply:

The HLM workscope does not include any extensive inclusion of parametric uncertainties, nor are any sensitivity analyses performed.

Comment:

For those tanks that are leakers, what are the uncertainties involved with estimated total leakage volume based on information obtained from monitor wells?

Reply:

This issue is outside the workscope of the HLM. As far as well know, the ranges given in Hanlon for these four tank leak estimates are all based on judgment about soil wetting with waste and not on calculations.

Comment:

With unknown condenser efficiency for reflux, the water and heat losses for that process would be difficult to accurately estimate. Is there really no data available for the reflux conditions?

Reply:

The only data that is available is the amount of condensate in kgal and often this data is incomplete.

Comment:

We consider it incorrect to assume cooling water temperature (e.g. 18 C) and that of the exit air were the same. The statement that, at times tank-specific coolers were required indicates that only 2% of the reflux volume was lost when larger losses could have easily occurred.

Reply:

We agree that these temperatures will in general be different. However, in that absence of data, inclusion of this temperature difference would introduce yet another parameter. Moreover, the amount of conductive heat loss to air is largely already accommodated in the 10 kW to air assumption. Therefore we will continue to neglect this temperature differential and lump the heat loss to air into a single parameter.

Comment:

Are all HLM evaporation and reflux rates based on the estimated tank heat load? Are tank cooling rates based on the measured tank temperatures? These bases are not clearly stated.

Reply:

All evaporation and reflux are based on tank heat load as stated in the text, p. 2 paragraph 2, "The HLM then reconciles this volume information with the evaporation rate that is expected based on the total heat load of each tank." There is no tank cooling rate in the HLM per se. There are two heat losses in the HLM: vaporization of water and thermal conduction to ventilated air and ground around the tank. The conductive loss to ground is indeed scaled with tank temperature. The measured tank temperature is used if it is available otherwise it is assumed as stated on p. 4 paragraph 3. We will make this more clear in the text.

Comment:

How do HLM results compare for tanks considered as non leakers?

Reply:

This will be added. In particular, we have evaluated SX-105, which is the only tank with lateral wells that has not leaked, and found that the HLM limit of leak detection is around 0.5 kgal/month, or a factor of three to six lower than any of the other four tanks in this study.

Comment:

We suggest that the HLM should be calibrated by performing calculations (with identical assumptions) for tanks where both the HLM and independent leak volumes can be (or have been derived). Considering the degree of uncertainty that appears to exist in this methodology, these comparisons (with comprehensive consideration of the uncertainties) appear necessary to address the question of whether this methodology can give meaningful results.

Reply:

The HLM is meant to be an independent method for estimating leaks for boiling waste tanks. We compare the HLM to these other estimates but do not calibrate the HLM leak to these other leak estimates. With regards to the amount of uncertainty in the HLM, we recognize that our methodology is very approximate. However, previous estimates are also very approximate. This is why there is a need for independent estimates. The comprehensive handling of uncertainties will require significantly more effort and therefore will need to wait.

Note that the HLM does not show leaks for SX-105, and does not show leaks for five other earlier cooling cycles for tanks SX-111 and SX-112. These other analyses support the interpretation of the HLM unaccounted volume losses as leaks.

Comment:

Why does the large turnover in tank volumes make determination of tank leaks so difficult?

Reply:

As stated in the text, large amounts of condensate invariably lead to material loss in the condenser system. This material loss will necessitate makeup volumes, and it will be difficult to differentiate this makeup volume from that required to replace material lost to the soil column as a leak.

Comment:

What are the uncertainties associated with the "unaccounted" volume losses?

Reply:

We are not going to do much with uncertainties at this stage of the model. However, we know that the smallest leak for the HLM is about 0.5 kgal/mo and this implies at least that amount of uncertainty with the leak estimate.

Comment:

How do results derived for other SX Farm tanks compare with those of the four tanks in the HLM?

Reply:

We have added one additional nonleaking tank (SX-105) to the HLM analysis for comparison. The remaining comparisons will require additional effort and therefore workscope. Generally, there are unaccounted losses in non-leaking tanks, and these losses define the limits of the HLM.

Comment:

How are sludge radionuclides such as Sr-90 partitioned compared to supernatant radionuclides such as Cs-137?

Reply:

When process waste is placed into a tank, both soluble and insoluble radionuclides are added together. Normally, the waste remains in each tank for at least a year to allow the short-lived radionuclides to decay. Thereafter, for supernatant transfers the HLM simply partitions the remaining heat equally between the supernatant and sludge.

Comment:

How does partitioning between sludge and supernatant affect the HLM? What uncertainties does the HLM partitioning assumptions introduce?

Reply:

Sludge/supernatant partitioning is only important for waste transfers out of the tank and therefore is most important in the tail of each tank’s cooling curve. This will be clarified more in text.

Comment:

How was evaporation modeled for this analysis?

Reply:

Evaporation was modeled very simply in the HLM. All heat above the conduction losses was removed by vaporization of water at 100 C.

Comment:

What is the uncertainty associated with estimating evaporative losses?

Reply:

Once again, the HLM analysis was not scoped to perform detailed uncertainties. However, we feel that the primary uncertainty in the evaporative loss is with our assumptions about heat load and conductive loss in the first place. Then, the heat of vaporization of water from waste is different from that we assumed, the temperature of the dome vapor will vary depending on many factors such as ambient air temperature, season of the year, time of the day, and ventilation rate. And then there are the losses of condensate from the system. These losses need to be made up with water additions.

Comment:

What fraction of "unaccounted volume increases" could be ascribed to uncertainties in information and methodology? What is "normal volume loss" and what are "inaccuracies of the HLM?" What are the criteria for deciding between "inaccuracies of the HLM" and "waste inventory that has been lost to the soil through a breach in the tank liner"?

Reply:

Uncertainties are beyond HLM scope. Normal loss is assumed to be that associated with makeup volumes and is not an HLM parameter at all as these losses are made up by water additions. There are no criteria for deciding between HLM inaccuracies and leak inventory. Leak inventory is defined by the leak parameters, which are in turn adjusted to eliminate unaccounted volume losses over some period.

Comment:

What are the uncertainties associated with the (calculated?) "evaporation rate" and unaccounted volume loss"?

Reply:

Uncertainties have not been estimated. The evaporation rate is indeed calculated.

Comment:

How would "a significant temperature decrease" cause "more evaporative heat loss"?

Reply:

A reported tank waste temperature decrease does not cause evaporative loss, it simply indicates it. That is, evaporative cooling will lower the waste temperature.

Comment:

It is very difficult to keep track of which values are measured, which are assumed, the basis for the assumptions, and the uncertainties associated with the assumptions and measured data.

Reply:

We will attempt to clarify this more in the methodology section. We do not have uncertainties for the data nor for our assumptions more than what is already stated in the report.

Comment:

Are evaporation rate and unaccounted volume loss measured or estimated by HLM?

Reply:

Evaporation rate is calculated based on heat load. Unaccounted volume loss is calculated based on reported tank volume, volume additions and removals, and evaporation. We will try to make this clearer in the methodology.

Comment:

It is stated that tanks that were cooled by evaporation are assumed to have been on a reflux condenser system. How can this assumption be confirmed?

Reply:

Since there are reports of condensate removal and water additions for these tanks, this is the basis for these assumptions. Presumably, there are logbooks associated with these operations that we have not yet found.

Comment:

Can details of the process be clarified? Is it possible to determine which tanks were on separate individual coolers and which were on combined ventilations systems? How will these uncertainties impact the results of the HLM?

Reply:

We have some more detailed information as stated in the text. That is all we have at this time.

Comment:

What is there to indicate that leak rates derived from HLM are meaningful?

Reply:

The scope of the HLM was to provide an independent estimate of leak volumes for these aging waste tanks. The finding of much larger than anticipated ground contamination around SX-108 certainly suggests that the previous leak estimates may have been too small. This question of how meaningful is the HLM is of course relevant to all leak estimates. How do we know any of these estimates are meaningful? In the end, both end up with a contaminated soil column, which we already know from various drywell observations.

Comment:

Is it possible to determine whether the condensate in SX-106 was used for makeup additions? Won’t this uncertainty affect the results given by the HLM? What is the basis for a 2% loss and how does it affect the HLM leak estimates?

Reply:

There are reported transactions from SX-106 to each of these tanks. Comments in Anderson list SX-106 as a condensate receiver. The 2% loss for makeup was used to illustrate a point and does not impact the HLM.

Comment:

Are several "periods of low reflux" used for determination? If not, extrapolation would appear to be uncertain. If more than one "period of low reflux" is used, what is the resulting standard deviation considering each period as resulting in one value?

Reply:

Each period of low reflux has volumes reported every quarter. Therefore, each quarter’s volume measurement contributes to the accumulated volume discrepancy. However, the overall driver in the HLM is the loss expected based on calculated evaporation based on heat load. Note that each of these tanks have already been assigned as leaking and therefore the HLM did not invent these leaks. The HLM simply interprets unaccounted volume losses and extrapolates them over time.

Comment:

What uncertainty is associated with the HLM assumptions of 20 kW conductive heat loss to ground and air?

Reply:

The traditional understanding of SST’s at Hanford has been that their passive heat loads were always limited to 40,000 Btu/hr or 10 kW. This kept waste in that tank below boiling (100 C) for the entire year. Ventilated DST’s are allowed twice that heatload at 20 kW for a similar reason. Therefore, the HLM simply uses these values as heat loss assumptions. These assumptions are most important in the tails of the tank waste cooling curves and this is of course the period for which the HLM is calibrated. Note that there are times when the waste cooling curve matches that calculated quite well. Otherwise, we have not evaluated this uncertainty of this nor its impact on the HLM leak estimates.

Comment:

We suggest a better description of assumed tank failure modes. Figures illustrating the scenarios considered would be helpful.

Reply:

The tank failure modes are not important for the HLM. We placed this in the report to provide some background.

Comment:

We suggest more information about the potential plume geometry.

Reply:

Discussion of plume effects was only meant to be qualitative and does not impact the HLM. Therefore, more discussion is not appropriate. There are other more appropriate references for plume geometry.

Comment:

The uncertainties associated with each assumption must be given if the validity of the HLM methodology is to be established.

Reply:

We agree that uncertainties are very important to completely validate HLM results. However, uncertainties are beyond the existing work scope. We nevertheless feel that there is still value to the HLM report despite these deficiencies.

Comment:

How were modes of failure and bottom bulges observed? What were the level changes associated with bottom bulges (and their disappearances) for SX Farm tanks?

Reply:

Bottom bulges were reported with sludge height measurement differences among risers. Also, bent thermocouple trees and other equipment were associated with bottom deformations. No liquid level changes have, to our knowledge, ever been attributed to bottom bulges. However, the volume changes that occurred for SX-108 in the 1959-61 period coincide with the reported bulge that then relaxed back into place.

Comment:

What is the depth of the drywells? Do any drywells extend horizontally beneath the tanks? A schematic figure showing the relation between waste level, assumed waste level, drywell locations, assumed plume boundary, assumed soil porosity, soil permeability, etc., would be extremely useful.

Reply:

Drywells range in depth from 35 to several hundred feet. We will add references to reports that detail these issues since none of these issues impacts the results of the HLM. There are drywells underneath tanks SX-107 through SX-115 that are termed laterals. They extend from caissons placed among the tanks with three laterals underneath each tank.

Comment:

We consider a better description for the reflux system and associated model assumption to be required. Since a large fraction of the HLM uncertainty may be associated with reflux systems, we suggest attempts be made to identify better information concerning the performance of the reflux systems.

It is assumed that the dew point of the condenser is 65 F. This requires either a refrigerated system or a very high performance cooler...Accurate knowledge of the ventilation system would appear to be critical...Do records for any of the tanks indicate local steam venting?

Reply:

We agree that better information would be desirable. However, we only have limited information and are necessarily limited by that until better information is available. More precise details of the reflux system may or may not affect the HLM results. The largest fraction of uncertainty is most likely associated with the tank heat load and conductive loss and not on condenser efficiency.

Comment:

What is the evidence supporting description of a leak by the two parameters, leak size and elevation? Why would leak rate be expected to be constant for any significant period of time for a given driving pressure?

Reply:

These leak parameters describe the simplest possible leak: a leak size that scales with hydraulic head and a leak elevation that defines the height of the leak within the tank. These parameters are used because they are the simplest. To start and stop a leak more than once requires yet another set of assumptions and it is not clear how we would assign those additional parameters except as we have for SX-112. In this case, a cooling period in between the two leak periods strongly suggest a two part leak.

Comment:

Paragraph 4 p. 5 seems to be contradictory. [This paragraph talks about self-sealing leaks.]

Reply:

Corrected.

Comment:

Why is the tank waste temperature assumed? Is it not known?

Reply:

We have limited tank waste temperature data. When it was available, we used it. Otherwise, it was assumed.

What is meant by chilled air that could either be vented or returned to each tank? Was this a closed system?

Reply:

Ventilation systems were (and still are ) operated in a partial recycle mode. The amount of return was often adjusted according to some particular criteria the details of which we have not found. For example, Tank C-106 is operated today with a chiller where around 60% of the vent volume is returned to the tank and only 40% is replaced by incoming air.

Comment:

Last paragraph p. 5 is not clear or helpful.

Reply:

This paragraph was meant to describe the gaps in the tank level data do not allow us to quantitatively ascribe level variations to waste evaporation. We will try to make is clearer by comparing periods where monthly instead of quarterly data were reported.

Comment:

The section describing the reflux model does not appear to support use of HLM to quantitatively describe a leak rate or volume.

Reply:

Once again, we feel that the heat load and conductive loss of the tank are the most important sources of uncertainty. There are many deficiencies in the HLM reflux model as pointed out in this section. We still feel that the tank heat load is likely the most important source of uncertainty.

Comment:

see 18), where need for uncertainties is reiterated.

Comment:

Why aren’t unaccounted volume gains considered due to accounting uncertainties? Why aren’t unaccounted volume losses also due to accounting uncertainties?

Reply:

Unaccounted volume gains are assigned to water additions by the HLM. Unaccounted losses above those expected from evaporation could correspondingly assigned to waste losses by transactions. The HLM is simply attempting to bound those unaccounted losses by reconciling evaporative loss with reported tank volume.

Comment:

Condensers are well known for having tubesheet failures. What is the basis for the assumption that a makeup of 2% of the reflux rate was appropriate or conservative. It would not take a very large leak in a condenser to add a significant volume of water to the tanks. Are there maintenance records relevant to this potential behavior available?

Reply:

The HLM does not depend on the 2% loss that is stated and this is only used for descriptive purposes. There are no records yet available for SX Farm operation with this kind of detail. However, 242-T and 242-S in-farm evaporators reported a 5% materials accountability criterion for their operation. Because the in-tank evaporation for SX Farm was a much more complex than the in-farm evaporators, this suggests that SX Farm condenser operation probably used a materials balance on the order of 5-10%.

Comment:

What comprises a bulge? How are they observed? Do bulges always have associated leaks? A schematic would be helpful.

Reply:

Bulges were associated with sudden changes in sludge weight measurements for particular risers. Therefore, their detection was dependent on the performance of a sludge height measurement performed close to the deformation. They also causes equipment in risers to become bent or deformed and all reported tank bottom bulges (SX-108, SX-113, A-105) have also been associated with leaks. It is therefore likely that there were other tank bottom bulges and that those bulges were never actually detected.

Comment:

Internal tank bottom deformations would have caused level changes. Did the HLM methodology include estimates for these factors? Could a relaxing bulge a the end of a high temperature period have been interpreted as a leak?

Reply:

Undoubtedly the tank bulges resulted in waste level changes. For example, the level changes for SX-108 in 1959-61 may have been at least partly due to a bottom bulge. A relaxing bulge could very well have been interpreted as a leak. However, the HLM does not include any compensation for level changes due to bottom bulges.

Comment:

What is the relation between lateral well activity and tank level stability (or leak rate) as a function of time?

Reply:

Within the HLM, there is no quantitative link between lateral well activity and tank leak rate. There is only the qualitative association that continued increase in lateral activity is associated with periods of time that each tank leaked waste by the HLM.

Comment:

What is the basis for the assumption that each leak rate by be ~50% lower than that stated within the HLM.

Reply:

This assumption was an attempt to get at the uncertainty in the tank heat load as well as the uncertainty in the leak "duty factor", or the period of time that a tank may actually have been leaking. This is not a quantitative uncertainty and is only a judgment at this point. Any quantitative description of uncertainty will require more effort and therefore expanded work scope.

Comment:

Notes on each tank do not appear to support the use of HLM for describing leak rate or volume.

Reply:

This information was meant to provide some background. The timing of the ground contamination and bottom bulges is important for the HLM, but otherwise the timing and leak rate are independent of past estimates.

Comment:

The HLM leaks are calibrated by reducing unaccounted volume losses by 80%. Why 80%?

Reply:

There is a variation of these unaccounted losses and therefore we did not feel justified in pushing it to 100%. An 80% value seemed reasonable in light of the other uncertainties present within the model. The filled diamonds on each plot show the remaining unaccounted volume losses for each tank (see comment 41).

Comment:

There does not appear to be a dependable basis for the leak rate and start and end times. Need to specify the assumptions and their uncertainties.

Reply:

The assumptions will be stated. As mentioned previously, uncertainties would require more work and are out of the scope of this effort.

Comment:

Factors that should be used to correct for soluble radionuclides are not clear.

Reply:

We will try to make them clearer.

Comment:

The description of the Redox process needs more detail. Need to state the bases for the earlier report and their uncertainties.

Reply:

We will add more detail about the HDW model bases. The uncertainties will have to wait.

Comment:

What is the basis for adding the amount of silica of 50 mol per ton of fuel processed?

Reply:

This is described in the HDW report and is related to that fact that silica shows up in waste analyses but was never added during processing. The primary known source of silica is from the bonding agent that was used in sintering fuel pellets. This addition is meant to reflect other silica sources that were unaccounted.

Comment:

We suggest a more complete description of how the spreadsheet was calculated.

Reply:

We will include these details.

Comment:

Figures require additional explanation. What is the significance of the filled diamonds? Why don’t figures seem to illustrate the HLM results?

Reply:

Figures do show HLM results as the greyed line. This is the calculated tank volume in the absence of a leak. The filled diamonds were meant to represent unaccounted volumes that remain since only ~80% of the unaccounted volumes were assigned to each leak.