Why is it known as a Killer Lake? What are Lake Kivu’s Dangers?

Gases, mainly CO₂, CH4 and H₂S, accumulate due to biogenesis in deep water up to 485 m. Gas in solution after hundreds of years is trending closer to saturation, contributing to Lake Kivu’s dangers. Lake saturation means the carrying capacity of dissolved gas is fully loaded. Any over-accumulation threatens an imminent limnic eruption.
The dissolved gas volume is almost as large as the 500 cubic km of the lake itself. This will mostly release like a giant geyser in less than a day. In Lake Kivu’s case this can be catastrophic to all life forms in the lake environment, where millions of people live today near the shores.
Panoramic photo of Mt Nyiragongo from Lake Kivu during wet season showing steam plume

Is Lake Kivu the Only Killer Lake?

No. An eruption, occurring at Lake Nyos in Cameroon in 1986, caused 1746 casualties. Another happended in Lake Monoun. As Lake Kivu is 10,000 times larger, millions of deaths follow if we do not prevent it.
So Lake Kivu’s dangers relate to its huge inventory of gas, so the consequences can be over 1000 times larger. A dense cloud of gas will fill the valley to a depth of 100 m or more. Anyone caught inside the cloud will be asphyxiated by the lack of oxygen, or more quickly poisoned by the released hydrogen sulphide content of 2000 ppm.

Can an Eruption of the Lake be Prevented?

Yes. The natural processes that create the biogas (methane) and co-products carbon dioxide and hydrogen sulphide have operated for thousands of years. It’s a near perfect environment for this to happen. So it’s not just about stopping gas generation, but about reversing any more accumulation to reduce the danger.
The “trigger” gas is methane as it’s much less soluble. Its 25 times higher partial pressure than CO₂ increases the eruption potential disproportionately. If one harvests methane, one can prevent an eruption. But the key to harvesting gas and preventing an eruption lies in how it’s done, in compliance with the rules. Doing extraction the wrong way can bring on a disaster even more quickly.

Is there a Right and Wrong Way to Harvest Methane?

We now know that there are several wrong ways, and perhaps only one right way. The difference lies in two main areas; (1) Is enough of the methane removed from the lake in the extraction process? (2) Does the process destroy or preserve the natural safety mechanism of the lake that keeps the gas contained in deep water?
We need to do the right thing, harvesting methane for commercial reasons to make the lake safe, in a way ensuring continued safety. The alternate to commercial production would cost billions and would never get funded.

When Will a Lake Eruption Happen and is it Predictable?

One can calculate when the lake would be certain to erupt simply, by following the trends. But the trend is changeable. We can even reverse the trend by our actions. Changes in nature can also vary the trend. Nature governs new gas formation; i.e. how fast or slow is like growing a crop.
Algal growth in the shallow waters depends on nutrients and oxygen. Wind and rain govern oxygen levels, with seasonal variation. But little -used agricultural fertilizers partly govern nutrient levels. Even more impactful is gas extraction bringing up deep water. It is N and P-rich, depositing it in the Biozone above 60m. Therefore algal growth rates can double or more.

What happens with the Carbon Dioxide when you Harvest Gas?

There is five times more volume of CO₂ in the lake than methane. About four times as much comes into the raw gas from older, simpler extraction methods. More advanced extraction techniques produce less. But in both cases we must water-wash the gas to remove CO₂.
The whole balance is important. It’s not only important to retain CO₂ deep in the lake, where it is safest, but it has a role in driving the extraction process. CO₂ washed out of the produced gas into shallow water would be better to return to depth instead.

What does Lake Kivu do for Climate Change? Is it a big Problem?

It can be a huge problem, but it has the potential to become a giant climate benefit. The big gas emission problem seems inevitable. The lake saturation is trending to cause a huge eruption in about 50 -70 years. An eruption can release as much GHG as the United States in a year, anywhere between 2-6 gigatons of carbon, on any single day.

How then can we benefit? It is prevention of what is otherwise going to happen, but with some great side benefits. Those benefits include providing cheaper, cleaner, carbon-negative energy to the region.

When can an eruption happen?

As it stands we seem destined to see it happen in plus/minus 50 years. But the known, unpredictably-timed potential for a new volcano or a seismic rift can pre-empt that scenario by decades. This scenario can trigger a major lake eruption any time from now. In one of the most active volcanic and seismic areas on the planet, this is an ever-present risk. However, the longer such an incident delays, the more severe an eruption could be as the gas inventory builds.

Is Climate Change the Main Reason to do the Projects?

Over time, the reason for and importance of developing Lake Kivu has changed. At first it was about filling an urgent need for power.

Recently it has become more focused on reversing climate change.
But overshadowing all of that, and potentially far more newsworthy, will be the humanitarian impact of a lake eruption. Millions of people could lose there lives on that same day. If caught in a toxic cloud in the dead of night, people would be defenceless.

With the best of safety alarms, with sirens warning people to get to high ground above the gas cloud, many people will not make it. Prevention is far better than any cure.

Where can one learn more detail about the lake?

The lake has become a special case in academia, dating back to the 1930s. It would not surprise if more PhDs and Masters degrees were awarded on this body of water than nearly every other one. It is a unique lake with uncommon characteristics. You will find many papers in the EAWAG faculty, in Kastanienbaum on Lake Lucerne in Switzerland.

Blog posts on this site describe and refer to more papers.
The expertise comes from academia, consulting and the engineering world contributed to the large body of knowledge and to the drafting of the rules for use of Lake Kivu. Eawag provided experts on the lake itself to the pro-bono Expert Group.

A recent paper published in Elsevier’s Science Direct was written by two members from COWI and Hydragas, in the Expert Advisory Group for Lake Kivu Development. It was published in January 2020. It speaks in great detail about Lake Kivu’s dangers and the solutions.

What are the Rewards for Developing the Gas Extraction Solution?

For the long lead time to developing a solution to Lake Kivu’s problems, that is a tough one. This is the province of scientists in the pursuit of knowledge, perhaps the publishing of one’s theory or invention. It can’t be for quick money. Most involved here earn little.

Now that’s where latecomers may have the advantage. They don’t have to spend millions of their own funds to pursue the ideas, experiment with solutions or to volunteer as an expert. In fact they clearly don’t have to commit the proverbial 10,000 hours to become an expert or the owner of a valid solution. They can wait, see who has it right, do some due diligence and then perhaps invest there in a fantastic opportunity. Lake Kivu’s dangers should not only be known by then, but their mitigation should be embedded in the solution

Can you put a value on the Value of Lake Kivu’s Development?

The best solution that comes out for the extraction process is, however, incredibly valuable. Beyond bragging rights of having resolved one of the world’s intractable problems, there are huge potential commercial rewards. Any company that gets to that position is therefore a candidate biotech unicorn.

The best solution can extract up to $60 billion in revenues from the lake, with >80% FCF and high returns. The efficiency of extraction and the recovery of available gas are vital in turning a non-profit into a super profits earner, through smart and world-leading technology.

Does that value include Carbon Credits?

No. At this time there are two methodologies for claiming carbon credits. First is the simple replacement of diesel with clean fuel, that counts for about 7.5 Mtpa carbon. The alternative biogas emits 1.19 Mtpa.

Secondly, but much more significant is the prevention of an eruption that would emit 1.9 Gigatons in a one-off event, calculated at 28 tons CO₂ per ton methane. Calculated at 87 t/t, the one-time carbon emission is 5.9 gigatons in a day. That number is close to the USA’s annual emissions of 6.9 gigatons in 2014. The value of these carbon emissions averted is huge; at just $30/ton, the value ranges from $60 – 175 Billion.


The Management Prescriptions Review Panel Asks: Is there a clear confirmation that the increase in exploitable methane mass in Lake Kivu is not discernible within the next 50 years?


(A)   In reply to the following comment: “On the matter of non-biogenic sources of methane and sudden inflow events and significant recharge of methane”: 
 
There is no credible scenario for a non-biogenic inflow of methane into the lake. Fossil methane is speculated to have formed in the Graben underlying lake Kivu. The proximity of hot volcanic zones under and to the north of the lake may compromise the potential of such a source of methane to exist for the long term. Also, fossil methane would almost always include ethane/propane/ butane in the mix. These are not present.
Biogenic production remains a more plausible explanation for “sudden inflow”, but only in a scenario where doubling the resource may take say ~100 years instead of 200.  It won’t be instantaneous.
On the matter of recharge: A better statement is that Schmid’s results, using not-fully-proven measuring techniques in the deep lake conditions, can neither confirm nor reject a significant recharge. What other credible source of methane exists?
If this is proposed as a conclusion, it is far-reaching and extremely important to properly verify. What is the alternate hypothesis on the origin of the methane? The measurement method may be subject to significant error. For example, if the error is -20%, the opposite conclusion could be drawn.
The conclusion raises questions and does not establish an incontrovertible conclusion yet. The techniques used were not yet convincingly calibrated for this purpose, for the mix of constituents or for the depth of resource. The measurement tool is however promising as a method and deserves consideration. But any gain in the lake methane content, or even unexplained reduction as may be proposed here, is dependent on other factors not mentioned here. For example:
·        The range of result in the Boehrer paper shows methane figures such as 18.0 +/- 1.0 mmol/l. at depth. That variability is +/- 5.5%. What is the comparable variance with other methods?
·        Is there a comparable commentary on the accuracy/variance of numbers back in 1975 and 2003? Could we have a combined variance of between 10-15% from then to now? If so, are any of the conclusions that can justifiably be drawn outside the potential error range and therefore statistically significant, or not? This is not explained.
 
(B)   The panel report states that: “The conclusion of this research is that the risk of a gas blow-out is not increasing, and additionally that also the lake’s methane resource is not renewable i.e. equivalent to a fossil fuel source.”  
 
If this is a conclusion, what is the alternate hypothesis on the origin of the methane? The measurement method may be subject to significant error. If the error is -20%, the opposite conclusion could be true. This is by no means conclusive or acceptable yet. Years more data are needed to provide proof for such slow change. The measurement error range is equivalent of 10-20 years of earlier assessments of rate of change.

(A)   The panel report further states that: “At some depth gases released from the riser flow are captured in a separator being a container at some specific depth (read hydrostatic pressure) were dissolved and free gases are in equilibrium.”  
This statement is neither good, nor scientifically correct, nor valid in any engineering sense. Gas release is not instant and therefore not in equilibrium in separators. Static systems over a long time reach equilibrium, but dynamic systems do not over a short time. That’s false.
 
(B)   The panel report states in this same section: “The natural ratio of methane to carbon dioxide in Lake Kivu appears to be too low to run biogas engines.”  
Appears? It is, at 20% methane, too low by a large margin. You can use that mix to put a fire out instead. 50% is close to the minimum combustible methane in gas for engines.
 
(C)  A further panel comment is “The less-favoured alternative by the expert group was the Density Gradient Draw-Down Method (GDM) though this method does appear as plan B in the MPs, see below.”  
In response, this “GDM” was not just “less-favoured”, but absolutely and comprehensively rejected by all other experts in the group. The GDM (which is NOT the same as Plan B) has a destructive impact on the natural stability structure of the lake. It also brings vast quantities of nutrients from deep water into shallow water. A principal reason for its rejection was the long-term instability of the lake. It comes from existing and (if and when new gas generation produced) fresh gas, that it would no longer be contained in the lower half of the lake. The same applies to all gas remaining after the first 50-year harvest.
 
(D)  The MPs are not clear whether this is per concession or per production unit and what should be the geographical dispersion of the production platforms.
Plan A1 requires smaller production units to reduce the impact of large volume extraction and re-injection flows on the gradient layers. A limit of 5 MWe in the output capacity was agreed at the 2008 and 2009 conferences of the experts. It was not an absolute figure but the best guidance at the time. The sensitivity of the gradient layers (260 m and 310 m) to large volumetric flows was the major concern. This is one of many contraventions by KivuWatt in its design.

(E) A further panel comment is “The less-favoured alternative by the expert group was the Density Gradient Draw-Down Method (GDM) though this method does appear as plan B in the MPs, see below.”
Tietze lobbied for this alternative on the grounds that he could “invent” a method of extracting all nutrients. He proposed that this could be part of the process of releasing the gas from the deep water. He could not offer any clue as to how this nutrient extraction idea could be achieved. He has not managed to do so in the ten years since. Therefore, to raise the GDM as a viable option is both dangerous and retrogressive science.
In the end Plan B turned out to be impossible to implement, as the required drop in density could not be achieved, without dilution causing the consequent “loss-of-use” of much of the gas in the URZ. Plan B should be disregarded as a viable option, as noted in the comment above, and deleted from the MPs.
(F) A footnote in the panel report says: “The MPs are not clear whether this is per concession or per production unit and what should be the geographical dispersion of the production platforms.”
The MPs do not imply that 5MWe is a limit on the size of a single production platform, nor for the capacity a concession. That is a mis-read. What it suggests is that a number of 5 MWe modules could be attached to a single barge/platform. i.e. Five such modules could produce 25 MWe and ten could produce 50 MWe. This is simple arithmetic, in case anyone missed the point. However, the spacing between individual 5 MWe production modules still had to be at least 100 m.
Kivuxxxx has placed four larger modules (80% over-sized) within 50 m of each other in their design. 50 m is too close. The only way this would have worked for them was to re-inject their degassed water into the PRZ, where the consequences are damage to the critical density gradient. It was ultimately a poor solution, perhaps their least worst available, but a destructive one at that.

What were the concerns of the Expert Working Group about short-circuiting?

(A) Hence, dilution of the URZ without energy production should be avoided to exploit this resource in the future. Yet, the gas pressures in the URZ are nearest to saturation limits, see Figure 2.4, though still at a safe level.
This may have been Kivuxxxx’s (mis)interpretation of the MTRs. The requirement clearly states that it should use the URZ first, as the zone more sensitive to poor re-injection practice. Symbion had a worse interpretation They assumed that they can re-inject degassed water to the URZ.
This litany of bad design practice may have included extraction modules being too large (>5MWe, Kivuxxxx & Symxxxx). Some were too close together (Kivuxxxx, Symxxxx), or unable to create a re-injection density sufficiently lower than that of the URZ (Kivuxxxx, Symxxxx, KP1). The depth had to be greater than at the centre of the 260 m Main Gradient. This density must be achieved by having the uppermost separator between 10-12 m below the lake’s surface. This requirement was not complied with by either Kivuxxxx or Symxxxx in their designs, with their top (only) separator at about 20-26 m below surface.
At first here may have been slight inaccuracies in the Chen-Millero re-injection density calculator provided by EAWAG. But any error was not nearly as gross as the 20+ m separator depth selection made by Kivuxxxx and Symxxxx (or perhaps Axxxxxx who carried out their engineering).

(B) The MPs do not specify the consequences of short-circuiting; we infer short-circuiting reduces (i) the methane concentration at the intake level as well as (ii) the upward flow in the risers because the gas lift is less powerful, reducing the upward methane flux even further.
Short circuiting results in the problem described in detail in Michel Halbwachs’ paper, quoted in this paper. The consequences are progressive dilution of the resource zone and premature failure of degassing. This problem was clearly identified in the development of the MPs. So was the need to position the top-most separator at a level that prevents mixing and hence this form of dilution.
So, of course short-circuiting would do that. Some things are obvious enough that the MPs do no need to spell them out in great detail. Short-circuiting from a poorly executed design was a warning in the guidance section of the MPs. It was only inserted as a rule to the extent that water had to re-stratify above the resource water through density difference. One should expect designers to have some logical capacity in their thinking.
Similarly, one expects that Halbwachs would understand that one has to separate re-injectate from the resource water. The separation works by re-injecting above the resource water at a density difference, as stated. This is why it is written up so in the MPs. The overriding need was not to place the re-injectate above the Main Gradient, due to much more serious consequences like destroying the Main Gradient. There was therefore no need to calculate the consequences of getting this wrong, writing a paper on the subject and blaming the MPs for making a mistake that it clearly warned against.

(C) Comment on p19: “The intake and the discharge in the same 40-50m thick depth range of the URZ may cause short-circuiting”

To repeat the above point: It may cause short circuiting if the density of the reinjection is not reduced sufficiently to create a sub-layer above the URZ. The control mechanism for densioty control is the depth of the uppermost (or single) separator used. KP1 has a single separator too deep to achieve the required density.
To repeat the above point: It may cause short circuiting if the density of the reinjection is not reduced sufficiently. The control mechanism is the depth of the uppermost (or single) separator used. KP1 has a single separator, which is too deep to achieve the required density.

(D) Given the suggested 20 years lifetime of Plan A1/A2 this example would involve 34-35 platforms with anchoring, risers, and pipelines distributed over Lake Kivu. This number of plants may be too extensive for commercial exploitation and investments and required spatial distribution over the lake (see also chapter 7 on concessions).

The question is built on a simplistic and incorrect assumption. Clusters of “5MWe” modules can be spaced out around a single platform. This solution is neither too extensive nor too expensive as implied. As described several comments above, it is a simple engineering decision. Use multiple, dispersed modules around a platform.
Kivuxxxx and Symxxxx both used multiple modules, but grouping them too close to achieve the needed effect of minimising negative impact on gradient layers. There are better designs available to fit the requirement, most likely even cheaper than Kivuxxxx’s design. They include appropriate anchoring for the modules.
It is a simple engineering decision to use multiple, dispersed modules around a platform. There are designs available to fit this spacing requirement. They are most likely much cheaper than Kivuxxxx’s design.

(E) Comment on p19: “No context is presented for allowing to change from Plan A1/A2 to Plan B.”
Plan A1/A2 were conceived at a time that the detailed density calculations to support their operation were uncertain. That density calculation by Chen-Millero was confirmed subsequently. The engineering needs to achieve the combination of A1 and A2 were confirmed.
Plan B was inserted as an alternate, potentially simpler plan to be tried treat the URZ and LRZ as a single zone. It required drawing resource water from the bottom of the LRZ and re-injecting degassed water above the top of the URZ, but below the centre-line of the Main Gradient. It proved unnecessary. Furthermore, the calculator proved that Plan B was much more difficult to design for, as it required the top separator to work under partial vacuum. No transition to Plan B is needed.
To continue from the above point: Plan B is problematic to operate and expensive to design, build and operate. One foresaw it as a standby option, but it proved to be of no value. One can demonstrate this in an engineering calculation, performed using the Chen-Millero calculator. It shows that the separator needs to have negative pressure to achieve the required density and restratification.
In the panel review document, Plan B is several times referred to being the same as KivuWatt’s design or Wuest’s PR1 concept. Neither is true because they both discharge degassed LRZ water into the PRZ. These options are highly damaging to the Main Gradient and must be banned.

(F) Comment on p19: “Subsequently, the MPs propose Plan B (Figure 2.8) in which riser water is taken from the LRZ but allowed to be re-injected in the now methane-poor URZ. The MPs define the latter circulation as Gradient Draw-Down Method (GDM) of the URZ-LRZ density gradient or interface, originally at 310m depth. Obviously, Plan B ends if short-circuiting occurs when the URZ-LRZ density gradient sinks below the riser inlet depth and methane concentrations drop to methane-poor URZ level, still leaving some deep volume of methane rich water not easily exploitable.”
There seems to be added confusion in the panel review on Plan B’s intent. Plan B was introduced to harvest the combined URZ and LRZ as one. This followed if there was a failure to harvest the zones independently by A1/A2. It is not the same as Tietze’s Gradient Drawdown Method, which is banned in the MPs. Although the intent of Plan B is repeated in the statement above, it gets mis-read and incorrectly confused with the GDM Method.
If one works with the Chen-Millero model, it can calculate the level of a separator vessel required to achieve a desired density. From a desired depth of water extraction, one can work out the installation depth. If the depth is above surface, i.e. to operate in a partial vacuum, one can see how Plan B becomes very difficult and expensive to implement. KivuWatt sought an easy way out and re-injected the water back into the PRZ. This has turned out to be disastrous for the Main Gradient, which is in the process of being weakened and ultimately will be destroyed by KivuWatt. This is why all the experts warned against KivuWatt’s plan and rejected the ESIA in July 2010.

Is there a clear confirmation that the increase in exploitable methane mass in Lake Kivu is not discernible within the next 50 years?

The papers and data sets from Boehrer (2018) and Schmid 2019 are not yet nearly sufficient to throw out all others. Their method used lacks well-proven validity over the depth range assessed as it was never previously calibrated for the depth or concentration of gas.
The method will require repeated surveys over multi-year intervals in order to establish either that there is a trend, or lack of any trend.
Further cross-referencing is important in order to establish Schmid’s hypothesis and claims of no-growth in the resource.
Uncertain measurement, including ~6% plus/minus data range uncertainty, along with any of a number of equipment failures in other measurement methods used from 1937 to 2004, could result in similar variances. i.e under-reporting or over-reporting is possible in any of the data series.
Similar measurement and calibration problems, including failure to account for sample leakage and temperature change, may well have compromised other tests. It is therefore way to soon to make or justify the conclusions put forward by Schmid, which have far-reaching consequences if correct.
Further, if accepted, the data suggest that methane is disappearing from the deeper resource zones. It was up to now common cause that the Main Gradient act as an impenetrable barrier to gas diffusion from below to above the gradient.
Only gas extraction or some form of organic consumption of methane could show otherwise. Neither seems likely nor is either proposed to account for a disappearance of methane. One or more set of measurements must be in error for such a thesis to stand, and Schmid is curiously selective on which data he proposes to ignore.
There are several scientific issues to consider, but the key references are:
• Every report on the gas concentration surveys from 1937 to 2019, which references are appended to the panel report and the MPs (up to 2004)
• The 1st Law of Thermodynamics (i.e. Matter can neither be created nor destroyed)
This finding of Schmid would effectively throw out all work done on Lake Kivu in the past and have all parties having to reset 83 years of past scientific work and measurement, his included. The physics, chemistry, microbiology, limnology, hydrology, geochemistry and other papers on gas extraction written on the subject would need to be withdrawn or re-written to support his hypothesis.
The MPs would need to be re-written to account for a static or declining methane load, and perhaps the carbon dioxide inventory too. The gas extraction methods actually still need to be the same to preserve the lake safety.
Or, we could check the measurements, past and present for errors.

How was the development of anchorage of KW barge?

If one evaluates the litany of engineering errors by developers to date, including:
• In 2006 the first methane extraction pilot plant REC experienced a riser breakage and sank (not correct, it was unseaworthy and sank)
• In 2008 the second methane extraction pilot plant REC experienced a structural/stress failure from poor design, was sinking. It was then ordered to be removed before it broke up and sank.
• KP1 experienced gas leakage, failing monitoring instruments and a riser that broke off, shutting down the plant for years.
• Kivuxxxx experienced problems with gas being over-saturated in wash-water, causing gas bubbling to the surface around the barge.
• The Kivuxxxx barge’s anchors began do drift in soft, spongy bottom strata, not allowed for in the anchor design, causing the barge to drift in high wind and current conditions.
• Kivuxxxx failing to design according to the MTRs 3 & 4, among others, causing problems with the stability and longevity of the Main Gradient layer.
Therefore, one can see a persistent trend of bad engineering, whether through incompetence or lack of know-how, but compounded by the impact of refusing to submit to (or a lack of imposing) the oversight of designs by competent experts. These requirements are set out in the 2009 MPs in great detail thus:
MAR1: Before applying for permission to proceed with the construction of gas extraction facilities, concessionaires must be able to thoroughly demonstrate that:
a) their plant designs and operational procedures will be in compliance with the provisions of this document; and
b) their Environmental Impact Assessments take these provisions into consideration.
Should existing facilities become non-compliant with an updated version of this document, the Bilateral Regulatory Authority will notify the operator of the non-compliance and the two parties will negotiate a mutually-agreed plan for bringing the facility into compliance.
MAR2: The locations of water intake and re-injection equipment will be approved by the Bilateral Regulatory Authority based on their depth, horizontal separation, and flow volumes.
MAR3: Prior to construction, design drawings must be submitted to the Bilateral Regulatory Authority for approval. This information will be kept confidential. Submitted designs must include at least all configurations of: underwater pipes, pumping systems, separators, gas lines, gas treatment facilities, gas buffer storage tanks, water mixing systems, compressors and blowers; power supply systems on off-shore barges; and gas flaring systems. All submerged materials of construction for underwater pipes and anchoring systems must be part of the documentation submitted.

Design drawings must include process flow diagrams (PFDs) and piping and instrumentation diagrams (P&IDs) prepared in accordance with ISO 10628. The design will have been subjected to the HAZOP process, and the HAZOP report must be submitted with the design drawings. This information will be kept confidential. The process flow diagrams (PFDs) shall as a minimum include the basic information as per section 4.2.1 plus items a (mass balance for all gases plus water), b, c, and d from section 4.2.2 of ISO 10628, and the P&IDs shall as a minimum contain the basic information as per section 4.3.1 plus items c, d, e, f and g from section 4.3.2.
MAR4: Relevant design data that will be reported at the design stage for all single extraction facilities, and that will be made public, include but are not limited to:
a) depth of all extraction and re-injection pipe openings;
b) design of the extraction and re-injection pipes: diameter, elasticity, heat conductivity, shape of pipe mouth to achieve internal mixing of re-injected water with surrounding water, ejector/diffuser, etc.;
c) design flow rates for all water extracted from or re-injected into the lake;
d) design flow rates including full mass balance, and expressed in SI Units (t/h or normal m3/h or km3/h etc.) for all gas streams produced during the extraction process;
e) the concentrations of methane, carbon dioxide and hydrogen sulphide in said water and gas streams, showing the extraction efficiencies and gas losses;
f) detailed and verifiable calculations of expected re-stratification levels for degassed water and for washing water; and
g) amounts of sellable power and of all internal power consumption in gas extraction.
Any subsequent changes to the above data must be submitted to the Bilateral Regulatory Authority for approval.

MAR5: The Bilateral Regulatory Authority has the right of access to gas extraction facilities for inspection at any time, and will provide facility operators with sufficient notice of such inspections. The operator has overall responsibility for safety at the facilities, and thus for granting access. The Bilateral Regulatory Authority personnel will have all necessary safety training as required by the operator for access to the offshore facilities. The operator will provide such training.
MAR6: At start-up of a new or modified gas extraction facility, a concessionaire must engage a qualified third party to carry out monitoring in the lake (e.g. salinity and temperature profile measurements) around the point of re-injection. This monitoring will, with sufficient reliability and precision, demonstrate the shape of the plumes of any re-injected water (degassed and washing water).

The purpose is to measure and report, from start-up until sufficient results have been reached, that there are no deviations from re-stratification levels as defined in this document. If necessary, adjustment of density control must take place followed by renewed monitoring until a satisfactory result has been obtained. The third party report will be submitted to the Bilateral Regulatory Authority.
MAR7: Operators of gas extraction facilities must report certain operation and monitoring data electronically to, and in a manner and frequency defined by, the Bilateral Regulatory Authority. Operators must be prepared to carry out automatic, online reporting if required by the Bilateral Regulatory Authority.
These data will be made public and used together with other data to develop better scientific understanding of the lake and guidance of extraction concession design, safety and operations.
Operator shall ensure that sample points for all major streams are installed, with suitable valving, for taking the necessary samples and for the Regulator’s appointed third party to take samples on request. Facilities shall be designed to allow carrying out tests with injection of tracers (through sample points).
Data to be reported include the following parameters:
a) Hourly averages of flow rates for all water extracted from and re-injected to the lake, plus the rates of produced gas or electrical power (MW) and cumulative production. Flow meters shall be calibrated once a year and copies of the calibration reports submitted to the Bilateral Regulatory Authority;
b) Monthly average values of flow rates for all gas streams produced by the extraction process, including mass balances showing the methane extraction efficiency and the relative carbon dioxide removal rate; and
c) Monthly average values of water temperature, conductivity and salinity, as well as concentrations of methane and of carbon dioxide, hydrogen sulphide and nitrogen in said water and gas streams, as well as calculated and/or measured densities of the re-injection water.

Other data may be added if concerns for the lake so require.
Once a month, a full set of water parameter analyses shall be made in an agreed laboratory on samples of extracted and re-injected water.
MAR8: Concessions should be limited to a time frame considered appropriate by the Bilateral Regulatory Authority. In considering an application for the renewal of a concession, the Bilateral Regulatory Authority should consider, among other things:
a) the overall gas reservoir management plan for the lake;
b) the availability of more efficient and sustainable gas extraction methods;
c) the potential need to change the extraction or re-injection depth and/or location; and
d) the potential need to change the carbon dioxide removal rate.
The Bilateral Regulatory Authority has the right to determine what technical renewal conditions will be required.
MAR9: A precondition for permission to start operation of any gas extraction facility is that the Bilateral Competent Authority or its representative has been given the opportunity to inspect all underwater piping and their fittings (as well as the materials they are made of) when assembled onshore but prior to installing these under water.


Frankly, Nothing More Needs to Be Said.
• Kivuxxxx failed on 9/9 MARs
• So did REC/Halbwachs
• So too did KP1, although it was largely built before the MPs were published.
To correct this: Repeat the same requirements (barring Plan B) in any version of the MPs, only with a stronger BRA and far harsher penalties for transgressions.


Under what conditions has KivuWatt been licensed?

From the statement on p22 of the panel review: “The set-up of Kivuxxxx is similar to Plan B in the MP, though with re-injection in the Potential Resource Zone (scenario PR1 in (Wüest et al., 2009)). Such a design would not comply with the MP, which calls first for Plan A1/A2 followed by Plan B.”
This statement shows that Kivuxxxx/Antxxxx did not understand (or chose not to apply) Plan B’s essential requirements. They modified Plan B by re-injecting into the PRZ. This utterly compromises the intent to protect the Main Gradient. Vaguely similar is not good enough; that’s as close as saying a bus is similar to a car as it has four wheels.
This design element is directly in conflict with one of the more important MTRs in the MPs. Kivuxxxx is unable to comply with many of the 14 MTRs, including the most important.
A Kivuxxxx-negotiated concession has stripped the GoR of any control over the mis-use of the resource. The lack of a BRA to enforce the MPs (MTRs and MARs) granted KivuWatt a sense of entitlement and immunity to sanction.
Concessional requirements have been abused to an extent illustrated by the lawsuit and arbitration case brought by KivuWatt against the GoR for issuing the 2009 MPs. The basis for the suit was “Change in Law”. This is despite the lack of any substantial difference between the Draft 2008 MPs document and the formally issued 2009 MPs on the MTRs and MARs that KivuWatt has flouted.
Kivuxxxx’s grievances on the above legal issue seem misplaced. They are trying to attach blame to others for what is a growing litany and engineering and design errors and failures.
Scientific background is substantiated by the MPs and the body of work that went into its formation.
The few known operational issues are largely anecdotal since Kivuxxxx publishes virtually none of the required data and reportage required under the MARs. For example:
• From the initial design data put forward in 2010 as “design information” to support Kivuxxxx’s ESIA application and application for MIGA support, the quality of information, its completeness and its accuracy was laughably inadequate. A few hand sketches, lacking even the basic engineering notation, were put forward for review.
• The absence of any design review or compliance review required by the above-copied MARs is a critical non-compliance.
• Their design, placed four large separators and their intake and re-injection pipes within a 50 m linear arrangement. It was against the guidance of the MPs.
• The report-back to the February 2011 Rubavu stakeholders conference, where the country manager of Kivuxxxx, reported that Kivuxxxx and its engineers “simply did not know how to design a re-injection system for degassed water to comply with the MPs.”
• Reportage of incidents, such as the aborted plant start-up in 2014/15. The resulting cloud of bubbles rising up under the platform from over-saturated wash water, was never reported.
• Their need to re-design and rebuild the gas washing system to correct the above design defect. Kivuxxx claimed this as a cost due to “change-of-law”, an absurd claim at best and fraudulent at worst.
• Kivuxxxx’s need to double-up on anchoring systems due to the failure of the original anchors to perform under the lake’s changeable seasonal wind and current conditions. This shows bad design practice.
• Most serious of all is the evidence from LKMP monitoring of the lake-wide weakening of the Main Gradient. Monitoring shows alarming evidence of the impact of degassed water re-injection into the PRZ. This contravention alone should require the immediate shut-down of their GEF.
• The overuse of resource, due to extremely poor process efficiency, plus the sole use of the resource volume in the band between 310 and 355 m depth, means that Kivuxxxx is accessing well below 50% of the available resource but overusing that band beyond its share. Arguably, Kivuxxxx uses or sequestrates between 3-5 times as much resource as the best available technology for the same net power output.


All the evidence above indicates that the MPs were substantially correct in requiring the technical compliances and administrative compliances. From my review of the MPs, there are many potential additions and one required deletion to the MTRs. The additions are written up in a joint paper written by Hirslund and Morkel, published in Elsevier’s Journal of African Earth Science, January 2020.
Perhaps a proper review would have thus dissuaded Kivuxxxx from building this GEF at all. Or it would have at least saved itself from so many embarrassingly incompetent design and engineering errors.
The former would have saved both the countries from expensive wastage of resource, not to mention an increase in the danger levels of the lake. The latter would have saved KivuWatt from tens of millions of dollars of wasted expenditure and years of delayed start-up.
As noted in the panel review document, the BRA was a necessary institution that should have been created to police the many infractions committed by Kivuxxxx. This need is even greater now, as the sense that developers can use political-legal mechanisms to flout the MPs seems to have become the “ruling precedent”.
We noted that Symxxxx proceeded in virtually the same manner in engineering and design of their proposed GEF. Most of the same design non-compliances were repeated by their engineers. Antares, also did similar poor work on the KivuWatt design. The design changes incorporated only modestly address deficiencies in Kivuxxxx’s GEF design.
 
Fortunately, REMA has seen fit to reject the ESIA put forward by Symxxxx/Shema, which has at least one design element that is worse than Kivuxxxx’s, with respect to re-injection.

Panel Review Statement: “Remove the scientific doubts between these conflicting views, record it and publish it.”

From the reported conflict on p22: Hirslund (2012) argues that in Lake Kivu chemoclines (constituent or dissolved-gas levels) or isopycnals (density levels) are moving upward by inflows to the deeper parts of the lake. These come from water sources with different salinities (citation from abstract Hirslund, 2012). From these inflows and consequent forced upward movement, Hirslund argues that local gas saturation increases. Thus there is an increase in the probability of gas eruption. Hirslund therefore proposes mitigating actions to limit the rise of chemoclines.
Schmid and Wüest (2012) object to these mitigations. While they acknowledge that observed thinning of the chemoclines from below is poorly understood, they hold that sub-aquatic sources maintain the chemoclines at their present levels.
Having been at the periphery of both of these arguments, I was part of a discussion on the merits of each. My initial take was that Schmid & Wüest were correct. They argue that the sub-lacustrine inflows would “shave off” tops of any rising chemoclines, keeping their levels constant despite thinning.
Moreover, in arguing this matter with Hirslund in favour of the view of Schmid & Wüest, Hirslund showed that the top fraction of a chemocline may be subject to this dilution/shaving effect, but that the body of the chemocline was not. DDL can explain the same phenomenon.
He also argued that sub-lacustrine inflows will settle at their own level, determined by the inflow’s density. Measurement from 1975 to 2004 showed that the gradients’ centre-line (or peak density change) was indeed migrating upwards. The impact of the moving centre-line is material over the long term mass balance changes.
However, much of the disagreement hasn’t been debated scientifically, but expressed more on a basis of personal animus and pride of authorship. It comes down to protecting one’s own thesis rather than doing a simple mass-balance to show that sub-lacustrine inflows will push up gradients.
With the combined effects of double diffusive layering, the sharpening the gradients, data also also shows that the peak of the gradient is perceptibly, and justifiably, rising. I now accept Hirslunds version of the mechanism and result, although less convenient than EAWAG’s. It creates lake management problems that will need long-term solutions.

To say that rising chemoclines is a myth and has no impact is wrong. Simple engineering mass balance puts that proposition down. The volume of water under each chemocline is increasing.
Operationally, rising chemoclines cause a growing and significant problem later on with extraction of gas. Rising chemoclines accelerate the rising partial pressure of gas where it presently peaks, just under the 260 m and 310 m chemoclines. This rising pressure persists even if, as Schmid 2019 postulates, no new gas is being generated.
Rising chemoclines will gradually, certainly not drastically, impact the Plan A1/A2 creation of new sub-layers in the URZ and PRZ. This is because with scaled-up production, the sub-layers will thicken faster than the rate of rise of the chemoclines.
In this context, mechanisms for rejecting degassed water to the Biozone need sober debate. Degassed water injected to the Biozone has multiple undesirable impacts on the Biozone ecology that must be measured against the safety impacts of rising partial pressures of gas at the top margins of the URZ and PRZ. But something will have to give over time and the alternatives need more study.
Extraction mechanisms proposed by Symbion/Shema do nothing to mitigate the problems of rising partial pressures at the present peak. This peak, at ~265-275 m depth, migrates upwards and grows to be more of a concern.
Ironically, KivuWatt’s method does slightly mitigate present rising partial pressure peaks at ~265-275 m depth and the 315-325 m as their mass redistribution to the PRZ slows the rate of rise of these 260 m and 310 m chemoclines. It does, however, still accelerate the rise of the 190m chemocline.
The 2009 MPs do not address the issue of rising chemoclines. Once the data and proven or disproven theories and mechanisms are settled, hopefully before another 5-10 years have passed, the matter can be addressed by qualified hydrologists added to the mix of experts.
MTRs 3 & 4 are affected by the discussion. A question of new MTRs to address the control of the level of key chemoclines must be addressed in concert with MTRs 3 & 4. While the LKMP and other researchers can help with providing more data, it is the wider set of expertise that will be equipped to address mitigation measures.
The matter is not one to be settled by Deltares and LKMP as they lack the requisite capabilities.

What arguments were used to prefer Plan A1/A2?

This is a short question that requires a detailed and wide-ranging answer to many issues, including:
(A) Halbwachs’ studies deserve the following response summary that we share in our evaluation – Reference is made to non-unanimity in the MPs expert group on the optimal extraction of methane.
A 5:1 majority was acceptable in the circumstances, particularly as Tietze did agree several times during detailed debates. He had agreed with our convincing technical arguments, data and models, only to renege days or weeks later. It was a regular pattern he exhibited over three years. In particular, throughout the 2009 finalisation of the MPs, Tietze agreed with the draft until the last day.
The rationale for the Expert Group rejecting the dissent of Tietze was overwhelming. His and Halbwachs’ views were rejected for good, reasoned arguments on the stability breakdown that would eventuate. They both argued for a method that would have destroyed the future stability of the lake by breaking up the important density structure.
A major safety problem reoccurs well after the completion of 50 years of gas harvesting. That is if gas content is regenerated after the initial gas harvest (as argued in this panel review). Their (Halbwachs and Tietze’s) harvest approach would have destroyed the existing density structure, resulting in a mixed layer taking up the entire lake. This results in a single density layer, like Lake Tanganyika.
Halbwachs issued his paper on the eve of the 2011 Gisenyi conference of stakeholders. The Expert Group nominated Morkel and Hirslund, at the request of Rwanda’s Minister of State for Energy – Dr Albert Butare, to prepare a reply to Halbwachs. The response, repeated here in full and unedited, was provided to him on 18 February 2011:

“To all Stakeholders in Lake Kivu Gas Extraction,
Re: Comments by Prof Halbwachs on Management Prescriptions (MPs) and in particular Plan A1, A2 and B Methods
Dear Colleagues,
As members of the International Expert Group we wish to respond to all parties addressed by the above correspondence, issued to most workshop attendees very shortly before it commenced. As such, we comment on the Halbwachs’ note and its attached YLec report from September 2009. Being addressed as late as it was, to almost all attendees, these were delivered as an “ambush” tactic to gain attention without allowing an opportunity of a timely rebuttal by the experts.
Nevertheless, we are obliged to properly consider the merits or otherwise of any communication on the MP’s, and to provide all those addressed with a sober evaluation and response. Both the Management Prescriptions and the conference have channels in place for the hearing of objections to the contents of the MP’s, neither of which was used by Prof Halbwachs. It is therefore principally to allay any concerns that could have been raised in the minds of other delegates that we reply as below.

The MPs have been conceived as Society’s tool to prevent future cataclysmic eruptions of the lake while extracting the gas. They are the only known tool to prevent this, and this is why the issue of adherence to the MPs is of exceptional importance. The comments by Prof Halbwachs do not qualify as sufficient reason to change this situation, as described below. If we were to follow these ideas of Prof Halbwachs, we would instead increase the certainty of many future eruptions of the lake.

“Pour tous ceux qui s’occupent de l’extraction du gaz du lac Kivu,
Note sur les commentaires par le professeur Halbwachs sur les prescriptions de gestion (les PG) de lac Kivu, et notamment les méthodes Plan A1, A2 et B
Chers collègues,
Comme membres du Groupe international d’experts nous souhaitons répondre à toutes les parties intéressées par la correspondance ci-dessus, délivrée aux participants de l’atelier à Gisenyi juste avant qu’il ne commence. En tant que tel, nous ne commentons que la note du Prof. Halbwachs et sa pièce jointe de Septembre 2009, s’adressant si tard à la totalité des participants de l’atelier, l’intention claire a été d‘attirer l’attention des participants d’un côté et de l’autre de ne pas laisser la possibilité d’un échange et surtout d’une réponse par les experts avant que l’atelier ne commence.
Néanmoins, nous sommes obligés de prendre dûment en compte le bien-fondé ou non de toute communication sur les PG, et à fournir à toutes questions qui ont été adressées, une évaluation et une réponse mesurées. Pour les PG ainsi que pour la conférence des moyens ont été mis en place pour permettre d’adresser les questions et objections sur le contenu des PG, ceux-ci n’ayant malheureusement pas été utilisés par le Professeur Halbwachs. Il nous semble donc absolument nécessaire d’apaiser les inquiétudes qui pourraient naître dans l’esprit des différents délégués et c’est justement dans ce but que nous répondons ci-dessous.
Les PG ont été conçus comme l’outil de la Société servant à prévenir de futures éruptions cataclysmiques du lac lors de l’extraction du gaz. Les PG sont les seuls outils connus pour éviter cela, et c’est pourquoi la question de l’adhésion aux PG est d’une importance capitale. Les commentaires du professeur Halbwachs ne changent en rien cette situation comme décrit ci-dessous. Si par hasard ces idées du Prof. Halbwachs étaient suivies, cela conduirait presque à une certitude de futures éruptions gazeuses du lac.


Background Issues
On the matter of their being included as part of the creation of the MPs, Prof Halbwachs and Dr Klaus Tietze are long-standing members of the Lake Kivu scientific community whose views are well circulated in publications. There are many parallels and commonalities in their viewpoints and opinions and, as such, one of the two was invited to be represented on the formative Expert Group in March 2007.
The broader scientific, consulting engineering, environmental and extractive technology community was similarly to be included in the group to make it properly qualified and representative. In addition, the group’s appointed members bring with them a broad set of additional skills which have markedly benefited the discussion and development of the MP’s. As such the group was properly and reasonably constituted and was not unduly biased in favour of or against any one particular set of opinions.
Further, in succeeding stages of the development of the Management Prescriptions, between October 2007 in Kastanienbaum and May 2009 where the final draft was prepared, open sessions of the experts were held and successive drafts issued for public comment. The opportunity to attend the May 2009 session was not taken up by Prof Halbwachs and comments on the draft MP’s were refused to be submitted on the “misguided” document on the grounds that he “knew better”. In the final issue of the MPs Dr Tietze was given a right to note his concerns over some elements of the MP’s and to reserve the right to prepare further research on his position and revert with proposed changes with justifications. To date none have been received for consideration.
The MP’s were issued as final in June 2009, and presented to the two governments in July 2009. After receiving copies of widely circulated negative comments on the MP’s, the Minister of State for Energy of Rwanda and the Deputy Minister of Hydrocarbons of the DRC both wrote to Prof Halbwachs to warn him against continued vilification of the MP’s and its authors through the press, including the writing of similar e-mails to the broader community. He was advised to deal with such matters through the relevant authorities, as appointed, and not in the public media.

Specific Objections
On the water re-injection method used after degassing, the YLec consultant’s competencies in fluid dynamics referred to are relevant to the argument on the predicted results of the extraction and re-injection piping and methods, but less relevant to the general safety of the lake. The public safety consideration always supersedes the economic priorities of either country, or any developers, in the order of precedence established in the guiding principles.
In the development of the Plan A1, A2 and Plan B as described in the MPs, these more detailed methods were instituted precisely because of the loss of gas recovery potential through dilution of the resource zones. It was well understood that dilution, caused by re-injection water distributing throughout the two mixed resource zones (URZ and LRZ), would render as much as 40-50% of the methane unrecoverable. Much of the research and discussion that preceded inclusion of these detailed extraction methods centred on their practicality and the relative improvement of gas recovery compared to the dilution method.

Hydraulic Evaluation of MP’s
In Copenhagen in both May 2008 and May 2009, two hydraulics specialists were mobilised by COWI to assist with evaluations. Before, during and after these sessions they advised on the 3-D impact of flows deriving from a variety of configurations of re-injection piping and the impact on flow regimes from the design of nozzles. These experts enjoy at least the same reputation as Yves Lecoffre of YLec, more especially in their hydrology experience of large unconfined bodies of water such as lakes rather than in piping systems.
These independent experts were called in to comment on and verify or dispute the opinions of persons in the expert group. Several of the Expert Group members have experience in the hydraulics of water bodies and flows. They had proposed certain configurations designed to vertically segregate the re-injected water from the underlying resource water in each zone. Detailed simulations were performed by these supplementary experts, based on the flow, density, and other data including the configurations of re-injection piping. Their results were used to determine the resultant shapes of plumes and re-injected water lenses in the lake, especially on the ratio of lens spread to thickness.
Where water of the precise density of the lower half of the relevant gradient layer was re-injected through a correct nozzle configuration, the re-injection lens was shown to stratify very widely, with a limited vertical penetration. With the correct parameters and designs, simulations showed lateral transport of 10 km or more being achieved with a vertical lens thickness of no more than a few metres, a ratio of the order of a thousand times. This spreading potential and segregation of the re-injected water is the basis of both Plan A1 and Plan A2 configurations.

Assumptions as Basis of Objections
The fundamental assumptions given to or used by YLec, gives rise to a serious bias in the results of their analysis. There are nine parameters that need to be well understood in a 3-D simulation, of which too few are mentioned in the analysis provided. Some are assumptions are made correctly, some incorrectly to the extent of being misleading, and too many are not even considered:
• The potential influence of the currents, even if for a transient condition, is discussed;
• YLec’s shape, configuration and flow parameters of re-injection nozzles are unlike any considered by the experts, and they do affect the analysis markedly;
• The assumption of re-injection density being the same as the mixed zone average density leads to the worst case result of zone dilution. Instead, this density should be assumed to being lower than the mixed zone, some 10 % lower than the difference between the mixed zone and the centre-line of the gradient above. This density is made possible by removal of the majority of the CO2 content of the resource water;
• Other parameters such as lens shape, the use of diffusers, selective withdrawal designs, flow rate, double-diffusivity effect on vertical and horizontal mixing should also be properly considered.
With the third point above, where the average density of the mixed-zone is assumed by Halbwachs to be the re-injection density, any resultant YLec simulation is not on a comparable basis with the Expert Group’s simulations. This assumption is wrong and designed to produce a misleading result.

Conclusions Drawn
The statement by Prof Halbwachs that his conclusions are final is rejected. The inherent bias, in both assumptions and modelling methodology, renders the work unacceptable as a basis for such conclusions or for recommending a change to the MP’s. Specifically:
The re-injected water can form a distinct lens, even becoming a distinct layer that forms above the mixed zone of the individual URZ or LRZ layers. During the conference, the simple example given of a cocktail that a barman can make with multiple coloured layers shows the alleged “impossibility” to be false. It’s a matter of skill and precision being demonstrated with fluids of very similar density.
That the dilution of a resource zone leads to a reduced extraction plant’s effectiveness is well understood. The extraction plans in the MP’s were drawn up precisely in order to minimise and delay this effect and then to be able to respond to the eventual drop in effectiveness of Plan A1 after some 20 years by switching to Plan B.
Given all the above, Prof Halbwachs’ conclusion that the description of the technology described in the MP’s as being “totally unrealistic” and “impossible to control” is (in itself) misleading, unfounded and without merit.

The issue of operating to Plan A1 and A2 will remain a concern for the designers and operators of gas extraction facilities, but the effort is necessary to maximise their commercial benefit from their respective concessions while preserving the density gradient layers.

Expert Conclusions
The clear, but unstated, inference from the Halbwachs report’s conclusions is that the MP’s should be based on the original legacy extraction method where the entire Resource Zone is displaced into the Intermediate Zone after degassing. This was the extraction method long-supported by Dr Tietze as well. However, this method unavoidably leads to weakening and, after 50 years methane harvesting, to the destruction of the key stability structures that have evolved to protect the lake for the past 1000 years from self-destruction.
A detailed set of quantitative evaluations and simulations of this methodology was analysed by EAWAG in 2007, with comparisons checked against the new proposed alternatives. The simulations of time-based variations to all parameters established the complex and long-term changes to the lake structure, and the dangers and effects of the total mixing of the lake’s zones. These included:
• The total loss after 50 years of protective density gradient layers that are essential to keep the future gas production retained safely, deep in the bottom half of the lake;
• The virtual certainty that after 50 years, the new gas accumulating will lift to just under the Biozone and will eventually result in frequent smaller eruptions of the lake;
• The injection of gas-bearing water of high gas partial pressure just below the Biozone which will, in time, increase the gas eruption risk with saturation levels up to 100%;
• The lifting of high nutrient concentrations to just below the Biozone will lead to significant impact on the biozone, including possible eutrophication. In the worst case this eutrophication leads to a dead lake unable to support sustainable fish life and can lead to seasonal releases of methane and hydrogen sulphide, in addition to carbon dioxide, from the surface as oxygen periodically depletes;
• The danger of the lake becoming a no-go zone for any surface vessels in the future is contemplated due to outbursts of gas originating just below the biozone. Frequent gas releases could lead to the shores of the lake becoming uninhabitable in the long term.
After considerable debate in the Expert Group on the need to consider the long-term after the 50-year harvest window, it was clear that the top two principles (Public Safety and Environmental Protection) would be considerably at risk from then onwards. It is quite obvious too that the social benefit from the lake, the third principle, would also be severely compromised for future generations.

It was therefore a clear and irrefutable decision by the experts that the Zone Mixing approach to lake harvesting had to be rejected completely in favour of the newer lake stability approach. The sacrifice of a small percentage of potential harvest in the medium-term (50 years) is not a high price to pay in the circumstances. These two divergent extraction methods are also completely incompatible so there is no room for accommodating both side-by-side on the lake at greater than experimental scale, as the chosen method relies on the stability and positional integrity of the density gradient layers to operate and perform.
We are not opposed to further hydrological and hydraulic evaluation and testing of the proposed extraction methods in the MP’s, and in fact encourage it. We would however insist on the proper use of both verifiable assumptions and models to ensure that a correct result is obtained.
Yours sincerely,
Dr Finn Hirslund P Eng
Philip Morkel Pr Eng.”


Our opinion expressed in the above letter remains substantially unaltered since 2011. No response or rebuttal to the above letter was received by the Expert Group nor arose until the process of this panel review commenced in 2019. The Halbwachs paper has been unaltered in the intervening years to cater for faulty assumptions made then.
Consistent with the opinion in the letter above, in response to Halbwachs contentions, no updates or corrected versions of his paper are published, to our knowledge. One assumes that correcting flawed assumptions and applying generally acceptable modelling techniques may have deflated the desired impact of the Halbwachs paper.
That said, if after nine years there has been no apparent advance by Halbwachs on improving the paper. The 2019/2020 re-issue of it to this review panel should equally questioned for validity of assumptions. So too the veracity of the results as a basis to argue against Plan A1/A2.

What arguments were used to prefer Plan A1/A2? (Continued)

(B) Comment on p25: “Re-injection in the (same) zone of extracted water creates dilution and should be prohibited, notably for Plan A1.”

This view was known at the time of the 2009 issue and had been discussed. Halbwachs failed to acknowledge that this hypothesis applied only to reinjection of degassed water at the same density of the resource zone (URZ).
But the stated requirement in the MPs was to reinject at density less than the URZ zone’s density. The difference must be sufficient to form a new sub-layer in the “knee” of the density profile. The density difference must be enough to remain separate from the URZ zone.
The detailed proceedings of discussions at the 2008 conference of the Expert Group were not published to provide as a reference. Parties to the conference, include several hydrology experts from COWI, with whom options were discussed. They include respondents to these questions. The collective opinion, after interrogation, was that the idea of re-injection to form a new sub-layer of the URZ or LRZ was accepted. The MPs were written accordingly. Dr Finn Hirslund must still have access to the same experts and can confirm the evaluation.

(C) Comment on p25: “A minimum efficiency in first stage separation should be imposed (75-80% is suggested.)” and “A maximum fraction of methane in re-injected water should be imposed (less than 20% is proposed)” and “A risk analysis of the evolution of a gaseous explosion is needed.” And “Thorough hydrodynamic simulations of the consequences of re-injection by an expert company are required (see below)”.

This imposition of these suggestion is not necessary or justifiable as a rule. Experts drafting the MPs were clear in saying that the process of extraction and gas scrubbing is a “black box”. The black box is the domain of the developer. However, all inputs and outputs from the “black box” are regulated, including their depths within the lake. It should stay that way otherwise the potential for innovation and design flexibility are lost.
What if there were ten separation stages? Would one stage have to do 80% of the work? That is silly and it is not the domain of the regulator to specify internal elements of the design. The regulator must, however, be convinced that the design is practical, adequately pilot-tested or full-scale demonstrated to allow it to be permitted.
The origins of these highlighted statements is not known, but they display a poor understanding of chemical engineering process design. The suggestions must be discarded as they make no sense and have no applicability to the broader options of plant design.


(D) Comment on p25: “The pie-shaped sub-division of the lake into concessions each having access to the deepest part of Lake Kivu located north-east of Idjwi Island is hardly feasible near the end (in time and depth) of exploitation. For access to the deepest parts gas-pipe lines should cross the lake to Rwandese or DRC shores with access to high-voltage electricity pylons respecting terrain and owners; a single concession should take over at the end.”

The pie-shape was the only way to give each concession access to the deepest water in each country. It is thus able to produce until the end of the concessions’ lifetime with the illustrated arrangement.
Safety and fairness of concession allocations were behind this choice. This alternative presented is unhelpful and barely comprehensible. The author of this suggestion is unknown, but he/she would be better served by withdrawing this statement.
There is no place for these “narrow interest, no context or self-serving” forms of comment or inclusion in the MPs. This is not in the domain of a regulator.
We must assume that a developer hires a competent engineer and that engineer can design to conform with regulated boundary conditions. They must perform engineering according to industry standards and with the necessary HAZOP or HAZID actions at the appropriate time. If any regulation should be applied to GEFs, it is this paragraph’s italicized sentence!
Further, these ideas apply to the type of legacy design which is inherently unable to operate within the bounds of the MPs. So these have no relevance to the discussion.

(E) Halbwachs’ study and more items on your p25 from the YLec Consultants paper have been quoted here with questions raised. They question items such as “The exploitable minimum…”, “A diffuser set-up…”, “Short-circuiting…”, “Single available study on gas lift…” etc.

Consistent with the overall opinion in the above letter response, here are some specifics that show the opinions not to useful, applicable, true etc. For example:
This 5 mol methane per m3 limit is highly dependent on the design efficiency of plant alternatives. It’s certainly true of designs for Kivuwatt, KP1 and probably Halbwachs. It is not true for multi-stage alternatives, which can operate where single-separator designs are unable to do so.
This diffuser set-up has been argued above already. The assumption made by Halbwachs and YLec are the same, that the density of reinjected water is the same as the zone. The MTR specifically calls for sufficient density difference to remain a separate sub-zone of the URZ. The problem is thus preventable and the conclusion unsupportable.
The argument, once again, only applies to the inherent weakness of designs. Examples include (1) single-separator designs, and (2) a failure to use the available controls of density. These ensure that densities are sufficiently different to remain unmixed in the URZ. The balance of this argument falls away as the base assumption is wrong.
The argument was simple and clear. The fundamental and best protection of the lake against eruption is to retain the current density structure. This must be retained forever, if possible, even if that requires long-term intervention to keep it stable. By re-injecting deep water into shallow water, the main gradient layer (chemocline) will weaken and then disappear. It may then take centuries to re-establish the stability structure and the safety it ensures.
If there is a theory that this will create a stable and static new Main Gradient, it is wishful thinking at best. Due to the stronger sub-lacustrine inflows above the current Main Gradient, this gradient will migrate upwards and disappear faster than before. The 190 m gradient is too high to retain much pressure (~19 bar vs ~26 bar) and therefore less gas storage.
As the gradient weakens, gas can migrate through it and will sharply increase gas partial pressure below the 190 m gradient. All benefits of stability will be rapidly lost. The lake will be a more dangerous place before the first harvest period of 50 years is complete. This theory is dangerous. KivuWatt is practicing this now at a smaller scale. It is already showing evidence of weakening the main gradient and compromising lake safety.
This Fig. 3.2 graphic has a fundamental weakness. It is an equilibrium-state calculation, not based on a dynamic state. Let me put this simply: If you open a beer and leave it so for 24 hours, most of the CO2 gas will have escaped. But it only approaches equilibrium 24 hours later. But the residence time in an extraction system is 2-3 minutes, over which time a far smaller fraction of a beer’s CO2 will escape.
Dynamic curves are very different. Likewise even the green curve, %CH4 recovered, is over-ambitious. 100% recovery is reached at much shallower depth in a dynamic system. It won’t get there with 2-3 minutes residence time. The curves are probably 50% over-stated. Halbwachs apparently lacks the data to show a dynamic curve. After all, his theory in 2003 was that it’s impossible to extract gas with a separator deeper than 20m. His own test-work failed to collect the relevant data. Hydragas is able to extract and separate gas at 65 m depth by contrast.
If the review panel has no objections to Halbwachs’ proposal, it has much to learn on this critical subject. This statement shows a worryingly low level of subject appreciation and knowledge. It cannot identify glaring and proven problems with Halbwachs’ proposal. Considering KivuWatt’s dangerous application of it, it is time to get some help. It’s really time to “go back to school” to learn not to make such dangerous statements.
The statement “Plan A1/A2 and Plan B require conditions of the hydrodynamic stability of Lake Kivu” is an example. It’s confused and out of touch. It is simplistic and uninformed, akin to one made in 2007 by Dr Tietze. His view was that “The only good lake is a dead lake. Take the gas out quickly and completely in 50 years. Then we are done.”
His view was dissected thoroughly and rejected for its many flaws by the Experts in 2007-2009. Fundamental to our rejections was that the world still needs to be safe well after 50 years. The lake will be there, biogenesis will be there (despite Schmid’s revelation that methane is fossil). People will be around the lake and they will be placed in mortal danger. The danger arises mainly by elimination of the best natural protection against gas eruptions. That protection is the density structure. It must be kept as-is.
On the statement “We conclude on (Halbwachs, 2011a,b,c ; Guillaume, 2009) as follows. The definition and function of concessions is of major concern. The reduced performance of gas-lift driven flow-rate in risers is not explicitly referred to in the MPs but of concern for determining the end of-exploration.”
Nothing stated above establishes any such rationale for your concern. Despite his theory of di-phasic flow, Halbwachs’ ability to create high-performing gas-lift is lagging the leaders. This italicised statement in the panel review shows a concerning bias. It provides little justification that stands up to inspection.
There seemingly exists an implicit distrust of the prior work Expert Group members. This is despite not having tested the arguments against Halbwachs’ paper. These are now provided. Two eminent COWI hydrologists did simulations and made presentations to the Experts in Copenhagen. They established that a stable re-injection lens could form in the given conditions. It would be a couple of metres deep and up to 20 km wide.
It is concerning again too that Halbwachs never sought to correct or update his paper for this panel review. It is nine years after it was issued and largely built on false assumptions. But he threw it again into the mix in a hope of not being found out for flawed assumptions. I suggest you set up a proper debate on the subject. It’s preferable to listening to one side only and pronouncing the above conclusion.
The Expert Group never issued published, formal working papers from its proceedings. Much of the work was consolidated by Hirslund in a 400-page treatise, shared among some members. This is still not published.
Halbwachs publicised his 2011 study to refute the MPs. But he made invalid assumptions in the paper as the basis of his argument. As a result, his paper is only true for a case mixing equal density discharges into the URZ. For the Plan A1 and supporting MTRs it is invalid.
The Expert Group spent almost two years, from 2007-09, dissecting Halbwachs and Tietze’s theories. They found them to be grossly flawed sometimes. So it is time-consuming to have to debunk them all over again. A major underlying flaw is their shared prediliction to destroy the lake’s density structure. This key structure is still widely acknowledged as the best natural protection against a lake eruption.
Hirslund has published a number of papers to describe his work and findings on Lake Kivu since 2009. I have taken the time to study them. Although few are in my field of expertise, they are a valuable source of insights into behaviours of Lake Kivu. One paper, jointly authored with me, is now published in Elsevier’s Journal of African Earth Science, January 2020.
https://www.sciencedirect.com/science/article/pii/S1464343X19303279. In this paper there is in-depth investigation of issues and ideas put forward by other authors in the field since 2009. There are also new ideas and findings on improving the safety of extraction.

(F) Comment on p28 (KivuWatt 2015-present): “However, to some degree the strategy followed by KivuWatt does comply with the MPs in the aspect of depth of re-injection and the diffuser designs which agree with the primary intention of the MPs.”
14 MTRs have vital requirements, serious requirements and good practice requirements. KiwuWatt is compliant with some less serious needs and non-compliant with the more serious.
But the MTRs are not a smorgasbord. They are mandatory, all 14 of them. Complying “to some degree” is not adequate. One good diffuser design does not excuse half a dozen non-compliances in other areas.
If anything, more MTRs are needed to tighten up KivuWatt’s further bad practices on a poorly engineered plant. Lawsuits initiated by KivuWatt are a political-legal argument against complying with the 2009 MPs. But these MPs were written six years before KivuWatt commissioned their platform.
Their lawsuit claims a “change-in-law”. KivuWatt reportedly preferred a “bootleg” 2008 early draft copy of the MPs, that a government official provided. But KivuWatt had been present, represented by their project technical director and ContourGlobal counsel in Copenhagen in 2009. Examination of the claims made in the lawsuit indicate that virtually all related to their poor engineering, rather than any change in law.
It may be legal practice to argue that prior law is “grandfathered” when it is updated. But this argument must be refuted when used to excuse placing millions of lives at greater risk. However, in this case there was no prior law.
Neither engineering nor construction were even started by time the MPs were first formally issued in June 2009. The relevant elements of the 2008 and 2009 MPs do not differ.
Should there be any access to view the proceeds of the KivuWatt vs GoR, there would be documentary evidence. Evidence both on behalf of the proponent and the defendant can be found there.

How would this in your view affect the current set of MPs or current operational practises?
(please specify the current number of the MPs that would possibly be affected)

It appears that the review panel scoured the earth for every negative sentiment, or every aggrieved proponent of debunked theories. This applies whether they were dealt with in the past or not. The same arguments are presented, without having addressed identified errors, false assumptions and unsubstantiated data. These prompted a previous debunking of the findings, conclusions and hypotheses in their work.
Curiously too, the work of KivuWatt has been presented here as a “good example”. In fact it has been more characterised by engineering and design flaws and lack of compliance. It demonstrates outright avoidance of compliance with technical and administrative rules. This lack of oversight provides clear evidence of no real knowledge and insight by the review panel. There is seemingly a willingness to believe anyone with a grievance.
As a result, there seems to be scarce constructive value emanating from this panel review. It is assembled with a suspicious mind-set, with most of the suspicion directed at the established experts and little directed at their critics.
This panel document did not identify the one MTR that was actually flawed. We acknowledge that MTR 5 on Plan B was flawed in its construct. But the panel document has falsely identified problems with other MTRs. These “issues” have not been established with any validity.

Why are conclusions offered in 3.1.5 on a series of contentions statements?

On p29 is a summary of Conclusions is offered in the panel report. Most of these are somewhat contentious and unproven, including:
3.1.5 Our conclusions on previous discussions on the MPs. Before we continue presenting our review, we summarize as follows the existing recommendations and amendments about the MPs. 1. The concerns and proposals presented by Halbwachs (2011a,b,c) and (Guillaume, 2009) are explicitly related to the methane extraction procedures of the MPs and point to voids in the MPs definition of concessions.
Dealt with before in 2011 and can be rejected again and again for the same reasons. The detailed letter of 2009 to Halbwachs stands, unanswered to date.
2. At least the part of Halbwachs’ proposal to re-inject de-gassed URZ water into the PRZ (Table 3.1) deserves closer attention.
This is dangerous, not improving on retrospectively approving KivuWatt’s bad engineering and failure to conform to the MPs. It deserves closer attention again to justify its rejection and continued banning.
3. The discussion by Tedesco et al. and Nayar invites risk assessment for cases where methane extraction is relevant or significant.
The opposite was true. MTRs in the MPs were made more difficult to comply with in order to prioritise safety of methane recovery. The principles in order were (1) people safety, (2) environmental safety, (3) community benefit and finally (4) developer benefits. We did not deviate and that is why KivuWatt, Halbwachs and Tietze all found the MPs difficult to comply with when issued. Their objections followed.
4. The methane extraction by KivuWatt since 2015 can serve as probe for the practical applicability of the MPs rules and guidelines for safety, inspection, on-site and nearfield monitoring, concession demarcation, legal instruments etc.
KivuWatt’s practice, both before and since the 2105 start-up, demonstrates gross errors in engineering. They show failure to comply with the MPs. They adopt their own interpretations of the MPs – neither grounded in the letter or intent of the MTRs and MARs.
KivuWatt failed to consult or take heed of the advice of experts. They didn’t even to attempt to understand the mandatory requirements for design. It is not a probe for practical applicability. Theirs is a lesson on how not to engineer, design, build and operate a gas extraction facility.
It could, however, be a lesson on how to stress-test the rules. They provide a check on authorities’ ability to verify and monitor circumvention of compliance.
5. Despite the exhibited knowledge of many aspects of Lake Kivu and methane extraction, we may appreciate Hirslund’s arguments more if his papers where (sic) more concise and his papers discuss the observations of, and with e.g. Ross and Sommer of EAWAG.
This statement is rich. The MPs are criticised for being too concise to explain the reasons behind all the Plan A1/A2 etc. When the rationale is presented by Hirslund in detail, his papers need to be more concise. Damned if you do, damned if you don’t.