Research results are presented below in five parts, corresponding to the five-part organization of CMI discussed above. Presentations at the Annual Meeting will elaborate on these summaries. Expansions of these presentations are found at the CMI website:

Figure 2


Carbon Capture and the Hydrogen Economy

Hydrogen Production from Fossil Fuels

Our carbon capture program is focusing on process and economic modeling of the co-production of H2 and electricity from fossil fuels. We are led to H2 as an energy carrier from simple carbon accounting: roughly half of fossil fuel carbon is dispersed to decentralized facilities before use (to boilers, furnaces, stoves, and vehicle engines) and half is used in centralized facilities (in electric power plants and very large industrial facilities like refineries and chemical plants). The unit costs for capturing carbon from centralized facilities will be far lower than for retrieving carbon once dispersed. Thus, if carbon dispersal can be avoided, then, economy-wide, twice as much carbon will be available for affordable capture.

Distributing only electricity where we now distribute carbon-based fuels is one way to avoid carbon dispersal. But probably cheaper and more versatile is an economy where, not one, but two non-carbon energy carriers are dispersed: H2 as well as electricity. Thus, we are focusing much of the carbon-capture program on understanding an energy system where, in centralized facilities, H2 and electricity are co-produced from fossil fuels and where, as well, by-product CO2 is captured. We are exploring, in particular, low-cost energy sources (coal, petroleum refining residuals), where the first step is oxygen-blown gasification. Natural gas systems will be investigated in CMI’s second year. See Figure 2.

Our strategy for entering this field has been to model a set of novel facilities, where one critical component is a scale-up of what is today only in the laboratory, but all other components are already commercial. The laboratory component is a hydrogen separation membrane reactor, in which the water-gas shift (WGS) reaction, CO+H2O  CO2+ H2, is promoted (and the equilibrium is shifted to the right) by the continuous extraction of the H2 reaction product. We are comparing these facilities to those that use conventional gas-separation systems – solvent absorption of H2S and CO2 and pressure-swing adsorption (PSA) for H2 separation.

There are several candidate membranes. We have focused on a non-porous intermediate-temperature (400-500°C) thin-film, sulfur-tolerant palladium-copper membrane. We (and other groups, including the BP-led Carbon Capture Project) find that membrane separation is promising, but no system yet identified is a clear winner. Both for the base-case plant with a membrane reactor and for the plant using conventional technology, the production cost of hydrogen in large plants, with sequestered CO2, is roughly 90 cents per gallon of gasoline-equivalent (lower heating value), or six dollars per gigajoule (higher heating value). Cost analysis is found in a powerpoint presentation at the CMI website. Related papers by Chiesa, Kreutz, and Williams, in various combinations, are in preparation.

Our physical modeling uses Aspen Plus and GS. Aspen Plus is widely used and versatile commercial software, ideal for creating a community of scholars who work to a common methodology. GS is process-modeling software created by our collaborators at the Politecnico di Milano to model advanced gas turbines and combined cycles. Using the two software programs in parallel gives us systems models that are superior to what either software could achieve on its own.

We have explored many variations in overall design and operation:

Co-sequestration of Sulfur (as H2S or SO2) along with CO2
The cost of CO2 sequestration may fall if other gases can be sequestered with the CO2. In the membrane system, co-sequestering SO2 with CO2 avoids the cost of flue gas desulfurization, but complicates removal of the highly acidic condensate from the raffinate turbine exhaust and drying and compressing the CO2/ SO2 mixture. In conventional technology systems, co-sequestering H2S with CO2 avoids the cost of the H2S absorption tower and the Claus plant for production of elemental sulfur. Co-sequestration may be an attractive option for pollutants other than sulfur, such as chlorine, nitrogen, and mercury. The CMI Carbon Storage Group is conducting a parallel investigation of the effects of co-sequestration on saline aquifer injection, geochemistry, and overall environmental risk.

Hydrogen Recovery Factor
Not surprisingly, the cost of hydrogen is a sensitive function of both the value of byproduct electricity and the hydrogen recovery factor (HRF, the ratio of separated H2 to CO + H2 in the syngas). Plant profitability has a broad maximum when the HRF is in the range of 70-90%. As long as there are markets for the byproduct electricity, higher values of CO-to-H2 conversion and HRF are not required, thus substantially relaxing the performance requirements of the membrane reactor.

Hydrogen Backpressure (i.e., Permeate Pressure)
For a fixed pressure upstream of the membrane, there are additional membrane area costs when the permeate pressure is increased, but there are reduced capital and electricity costs because less compression of the separated hydrogen is required to reach the pressures needed to pipe the hydrogen from the plant to consumers. Optimal backpressures are found to range from 1.5 to 20 bar, depending upon assumptions about the cost and H2 permeance of the membrane.

Raffinate Turbine Characteristics
The syngas, after hydrogen extraction, produces electricity in a “raffinate” turbine. We find improved efficiency but little cost savings for a raffinate turbine with an open steam circuit for blade cooling, relative to uncooled blades. The same holds for operating the raffinate turbine at a lower exhaust pressure.

Membrane Reactor Operating Temperature
We find that the system efficiency falls when the operating temperature of the membrane reactor is increased above the base case (base case = 475°C), for a fixed steam-to-carbon ratio, because the base case system is well heat- integrated, and raising the temperature requires either syngas combustion or expensive syngas recuperators.

Membrane Reactor Integration
At intermediate temperatures, the WGS reaction occurs so rapidly that the membrane reactor effectively acts as an upstream adiabatic WGS reactor, in which most of the CO is converted to H2, followed by a membrane permeator with only a slight degree of WGS reaction. In order to avoid significant thermal gradients and catalyst changes in the delicate membrane reactor, we have added a dedicated high-temperature WGS reactor so that most of the WGS reaction occurs upstream of the membrane reactor.

Collaboration with Tsinghua University
In the second half of CMI’s first year, BP created a research program at Tsinghua University (Beijing) and Dalian Institute of Chemical Physics of the Chinese Academy of Sciences. Some China funds are to be used for work with Princeton, acknowledging a longstanding Princeton-Tsinghua collaboration. Some CMI funds were reprogrammed to respond to this opportunity.

In the past few months, as a result of visits back and forth, joint work has begun on polygeneration strategies – the co-production of clean liquid and gaseous fuels, electricity, and chemicals from gasified coal. With the goal of analyzing the component modules of alternative polygeneration plants, our Tsinghua partners have begun using Aspen Plus to create libraries of energy conversion devices.

Polygeneration strategies, in the short term, should produce cost-competitive clean energy, with major benefits for local air quality. Moreover, these strategies will foster the key enabling technologies for a transition over the longer term to a H2/electricity economy with CO2 sequestration.

Infrastructure for Transmission of CO2 and H2
Also launched in the second half of CMI’s first year is an “infrastructure project.” Two new infrastructures of transmission pipelines, one for H2 and one for CO2, are implicit in the overall concept CMI is exploring. The project is documenting how relative costs of production, transmission and distribution, and preparation for use or disposal depend on scale, use of trunk lines, pipeline pressures, value at point of use or disposal, and co-sequestration specifications.

Hydrogen Combustion
CMI’s H2 combustion program has three main thrusts: 1) performance optimization of H2 combustion in heat and power generation, 2) control of fire and explosion hazards, and 3) use of H2 /hydrocarbon mixtures. Specific issues of interest are high-pressure chemical kinetics and flame dynamics of relevance to internal combustion engines, ultra-lean combustion for enhanced efficiency and reduced NOx formation, self-turbulization and transition to detonation, control of explosion hazards, and benefits from adding hydrocarbons to H2 and vice versa.

Experiments on the outward propagation of centrally ignited flame spheres of lean mixtures shed light on the onset of turbulence and the path to detonation. Hydrocarbons and H2 differ: while the hydrocarbon flame surface remains smooth and propagates steadily, the H2 flame surface rapidly wrinkles and accelerates. We are exploring whether addition of hydrocarbons to H2 suppresses the development of flame wrinkles and the transition to detonation.

New computations support the idea that hydrocarbon addition can significantly reduce the propensity of H2 to sustain detonation. In simulations, the induction length of detonation waves of mixtures of H2 and hydrocarbons is substantially increased with hydrocarbon addition.


Carbon Storage in Deep Saline Aquifers

During the first year of the CMI project, the Carbon Storage Group (“Aquifer Group”) identified important practical questions regarding CO2 storage in geologic formations, formulated relevant science questions associated with these practical problems, assembled personnel to address these issues, and developed a set of specific objectives and tasks to solve the problems. Overall, two broad issues are being studied: 1) environmental effects and safety, and 2) overall storage capacity available in deep saline aquifers. The first will determine whether geologic storage is environmentally acceptable. The second will determine whether it can solve the carbon problem. An overview of each of the major research thrust areas is summarized below.

Development and Applications of a Parallel, Multi-Physics Simulator
The Aquifers Group decided at the outset of the project to develop in-house codes to simulate CO2 flow physics, geochemistry, and geomechanics. We have expanded the capabilities of the finite element code Dynaflow, developed over the past 20 years to solve equations of fluid flow and geomechanics in porous media (see The code is parallelized and uses modern numerical techniques that include local refinement and highly efficient solvers. Into this code we have added two-phase flow modules and equations of state appropriate for CO2-brine systems. These equations of state span the temperature and pressure ranges that allow us to simulate CO2 migration from the deep injection zone to near-surface regions associated with leakage. We are adding geochemistry modules to this simulator.

Potential Escape Pathways and Associated Environmental Effects
One of the central tenets of our work is that leaks will occur, and that we need to be able to deal with them in an environmentally acceptable way. Our approach includes the following steps: (1) identify all possible escape pathways, (2) identify and quantify associated environmental consequences of leaks, (3) quantify leakage probabilities, and (4) estimate environmental risks based on the quantitative information from the previous steps.

Most of our efforts in the first year have focused on identification and quantification of environmental effects of leaks in shallow subsurface zones. We have used a contaminant transport simulator, which includes a detailed geochemistry module, to predict the effects of CO2 leakage on the water quality of shallow freshwater aquifers. Our simulations show the extent to which a CO2 leak reduces the pH of the aquifer water and thereby mobilizes potential contaminants such as metals. One specific case of CO2 leakage into an aquifer containing the mineral, galena, indicates dissolved concentrations of lead above drinking water standards. We are also developing density-dependent simulators to examine distributions of CO2 in unsaturated soils. And we have initiated an experimental program to investigate the soil geochemistry of CO2-rich environments, using soil samples from Mammoth Mountain, California, the site of a natural CO2 leak, where extensive vegetation death has occurred.

As part of this overall effort, we have written a manuscript (Bruant et al., 2001) in which we lay out the case for safe disposal of CO2 in deep saline aquifers. The manuscript is available at the CMI website.

Laboratory and Field Investigations of CO2 Interactions with Geological Materials
We have initiated two laboratory measurement programs aimed at quantification of CO2-brine-rock geochemical interactions. The first is a laboratory, which has now been completed, in which reaction rates can be measures at the core (centimeter) scale. This facility can accommodate high pressures and temperatures, and will allow us to determine reaction coefficients under conditions expected in deep saline aquifers, as well as those expected in shallow freshwater aquifers. A second laboratory has been constructed in which molecular-scale reaction details can be studied at the fluid-mineral surface. Most of the first year was spent designing and building these laboratories. We expect to run our first experiments early in 2002.

Identification and Analysis of Candidate Field Sites
While many of the research efforts at Princeton involve computational studies and laboratory experiments, these efforts ultimately will be driven by data from field sites. Initial investigations have identified the Alberta Basin, in the Province of Alberta, as a good test case for our initial computational and experimental efforts. Because Alberta requires all information obtained during drilling to be turned over to the Alberta Energy and Utilities Board (EUB), all information is publicly available. Cores extracted from exploration wells must also be turned over to the EUB, which maintains a Core Research Center where cores from more than 50,000 wells, and drill cuttings from more than 100,000 wells, are stored and are publicly available. The Alberta Basin is also attractive as a test basin because there are more than 30 ongoing acid gas injection sites within the basin. These injections involve mixtures of CO2, H2S, and perhaps other gases.

Figure 3: Mass of CO2 per unit volume as a function of radial distance and depth for continuous injection of CO2 at 20 kg/s, and injection of recirculated brine above the confining layer at 40 kg/s. Uniform layering of material properties is assumed, with permeabilities of 100 mD in the receptor formation, 0.1 mD in the confining layer, and 100 mD in the unit above the confining layer.

We plan to use at least one of the acid-gas injection sites for detailed simulation studies. We also plan to use samples provided by the Core Research Center to perform some of our geochemical experiments. We have a visit to Calgary and Edmonton planned for late January to arrange for core retrieval and to plan details of the proposed experiments. Earlier visits have established a good working relationship between Princeton and the EUB, and a formal agreement is in place, structured formally to be independent of CMI.

Hazardous Waste Regulation and the CO2 Regulatory Structure
To understand the regulatory framework that already exists for deep fluid injections, we have studied the regulatory structure for hazardous waste injection into deep geologic formations. In a visit to EPA offices in Texas, we read permit applications to assess the information included in those petitions. See the CMI website for details.

Brine Management and Control of CO2 Migration
If massive amounts of CO2 are to be injected into deep saline aquifers, a number of questions arise regarding behavior, migration, and possible management of the displaced brine. We are asking whether the brine can be manipulated to improve control of CO2 migration and to decrease probability of leakage. To answer this question, we have conducted a simulation study in which supercritical CO2 is injected into a formation, and resident brine is manipulated (withdrawn and reinjected) to provide more favorable pressure gradients. We find reductions in upward CO2 fluxes, leading to suggestions for possible injection designs to minimize CO2 leakage. See Figure 3. Details may be found in Guswa and Celia (2001), available at the CMI website.

New Initiative: CO2 Transport through Well Cements
Analysis of potential escape pathways for injected CO2 makes clear that escape via existing wells may be one of the major pathways. In the Alberta Basin, more than 300,000 oil and gas wells have been drilled. Similar situations exist in West Texas and other locations. If CO2 is to be injected into such formations, careful analysis of CO2 transport through these boreholes must be performed. Information is required about historical well completion and abandonment practices, the physics and chemistry of CO2 transport in cements, degradation of cements as a function of time, and the effects of additional components like H2S on cement reactions and properties. To address these issues, we have included a new faculty member in our group, George Scherer, a materials scientist knowledgeable about cements. We are initiating a program of numerical simulation and laboratory work to investigate transport of CO2 through well cements and other materials of various vintages used to complete and plug wells.


Carbon Cycle Science

The Carbon Science component of the CMI has three functions:

  1. It provides the carbon cycle science and climatology necessary to determine how effectively a program of mitigation will control atmospheric CO2 and reduce global warming. How much sequestration is enough?
  2. It provides assessments and modeling studies of alternatives to geological sequestration. The idea here is to prevent the CMI from discarding a technology too early, and to keep abreast of the activities of other carbon mitigation groups. We cannot fully assess the potential of geological sequestration without understanding the alternatives.
  3. It leverages the order-of-magnitude larger carbon and climate research program at Princeton and the Geophysical Fluid Dynamics Laboratory to reduce uncertainty about greenhouse warming and to anticipate new developments that may shape public opinion.


How Much Sequestration is Enough?

At global scales, CO2 leaks from geological reservoirs might be large enough to make sequestration counterproductive. For example, a trillion tons of carbon stored in aquifers that leak 1% per year would produce 10 billion tons of emissions annually, which is more than current global emissions of 7 billion tons. How small must leaks be if we are to solve a large part of the global warming problem by geological sequestration? The CMI calculated constraints on reservoir leakage using models of carbon storage reservoirs and global models of natural carbon sinks (Princeton ocean and terrestrial biosphere models). Future fossil fuel consumption was assumed to be at a level that would cause substantial warming in the absence of sequestration (the 1996 IPCC S750 scenario, in which atmospheric CO2 eventually reaches 750 ppm). The study then calculated the sequestration and leakage limits that would produce a maximum CO2 concentration of 450 or 550 ppm (the IPCC S450 and S550 scenarios).

The surprising result is that leakage limits are much less severe than expected because of heterogeneity among reservoirs. The increased retention of low-leakage reservoirs more than compensates for the increased losses from high-leakage reservoirs. For example, the reduction from 750 to 450 ppm would require a mean leakage rate less than 0.1% per year if all reservoirs were identical. In contrast, if leakage rates were sufficiently variable across reservoirs, then the reduction from 750 to 450 ppm would be possible even with a mean leakage rate of 1% per year or more. This study makes the scientific case for geological sequestration considerably stronger. See our website for more information.

Oceanic Iron Fertilization and Direct Injection
In the past year we have examined how fertilization that is patchy in space and time affects carbon cycling. Definitive answers require better understanding of the behavior of particles produced by fertilization, after they have sunk below the surface layer of the ocean. However, several of our preliminary results raise significant questions about the verifiability and consequences of micronutrient fertilization as a means of sequestering carbon:

  1. In our models, only 2-10% of the additional particulate carbon flux to the deep ocean resulting from fertilization is supplied from the atmosphere.
  2. Fertilization produces a complicated pattern of air-sea carbon fluxes over a wide area; these perturbation fluxes are small relative to natural background fluxes, making direct verification of carbon uptake impractical.
  3. When carbon is exported to great depth, the concomitant export of nutrients results in a significant decrease in subsequent export production; this may, over a period of a century, have a substantial negative economic impact on fisheries. See the CMI website for additional information.
Figure 4: Shipboard Automated Atmospheric Oxygen Sampler

In our original proposal, our overall assessment of direct oceanic injection of CO2 was that: 1) Direct injection is an efficient way to sequester CO2 if the injection is made deep enough. 2) The economic cost is probably much higher than the cost of geologic sequestration. 3) The ecological risks of direct injection are potentially severe. (These risks are beginning to be recognized by the scientific community; see Science, Oct. 12, 2001.) We are now participating in an international project to compare models of direct oceanic injection of CO2. This project is resulting in much improved estimates of the sequestration efficiency for many injection sites and a much better understanding of the uncertainties in these estimates. (See the Figures on direct oceanic injection in the carbon science power-point summary at the CMI website.)

Reducing Fundamental Uncertainty
The CMI also allocates the equivalent of four researchers (three post-docs and one technician) to critical emerging areas of basic research in four areas:

High-Precision Measurement of Atmospheric Oxygen to Quantify Natural Carbon Sources and Sinks
M. Bender’s group has completed construction of two automated oxygen samplers. See Figure 4. One is installed on Ka’Imimoana, a NOAA research ship operating in the equatorial Pacific. If we are to deploy these instruments on BP ships, we will require improved cooperation from BP shipping.

Improved Methods to Estimate CO2 Sources and Sinks on Land and in the Oceans
Critical to improving the accuracy of estimates of the magnitudes of carbon sources and sinks is better knowledge of the movement of CO2 in the atmosphere and/or oceans. We are developing methods to provide such knowledge, including methods that depend on Bender’s oxygen and argon measurements. The research has a potential application in locating leaks from geologic storage reservoirs.

The Link between Ice Ages and Atmospheric CO2 
We cannot be confident in our capacity to predict the future of atmospheric CO2 or its climatic effects until our models can explain what has occurred in the past. The history of CO2 levels during recent ice ages and interglacial periods, revealed by ice cores, provides a natural test case: atmospheric CO2 was lower during ice ages than during interglacial periods, thus causing additional cooling during ice ages. A central unanswered question is: What caused the concentration of CO2 to change? The same question is crucial to understanding the feedbacks associated with anthropogenic CO2 additions. Using the stable isotopes of nitrogen in deep sea sediments to study the history of ocean nutrient cycling, Sigman and his colleagues have developed the hypothesis that during ice ages the polar ocean is stratified, both in the Antarctic and in the Subarctic Pacific, preventing the release of biologically sequestered carbon from the deep ocean and thus lowering the concentration of CO2 in the atmosphere. Sigman’s group is currently testing this hypothesis with new data and evaluating physical mechanisms for the link between climate cooling and polar stratification. The urgency of this work is enhanced by the fact that ocean general circulation models, a widely used tool for ocean climate prediction, do not appear to confirm a causal connection between global cooling and polar stratification. If Sigman’s hypothesis is correct, then the simulation of polar processes by these models is functionally unreliable.

Climate Change and Land Use
We have completed the development of two models that predict the effects of land use on the weather. A preliminary result is that land use change in the U.S. over the past 300 years has cooled the climate. Thus, the observed warming in the U.S. would have been greater without land use change. We are currently sifting among the possible alternative explanations for this effect. Also, the GFDL/Princeton team is on track to complete the development of its Earth system model (atmosphere-ocean-land) by the end of 2002. We will thus have an exciting new tool to investigate the complex feedbacks that govern the carbon and climate system.

More information about these projects is found at the carbon science section of the CMI website.


Carbon Policy

During the last year, the Carbon Policy Group, the newest group of CMI, has focused on three interrelated projects. First, we developed a general economic framework to analyze the economically optimal timing and extent of CO2 sequestration. The analysis suggests that the technological option of CO2 sequestration reduces the economically optimal CO2 concentrations considerably. The extent of CO2 sequestration is affected by several factors, most importantly the marginal abatement and sequestration costs, the rate of cost reduction with increased installed capacity, and the economic damages imposed by climatic change.

The economic damages imposed by climatic change are important, because larger climatic damage increases the optimal abatement cost at the margin. Expected climate damage, however, is very uncertain. This uncertainty motivates the second research project, the analysis of the economically optimal response to climate thresholds. Damage from gradual and slow climate change is usually estimated to be relatively low. Damage from abrupt climate change may be more important, because its surprising nature and speed are obstacles to adaptation. We focus on a particular (some would argue, the most likely and relevant) climate threshold, a collapse of the ocean circulation system.

Our third research project is an exploration of a promising alternative to the Kyoto-style cap and trade approach to greenhouse gas control. Specifically, we develop a “No Cap, But Trade (NCBT)” approach. The proposed alternative has two main advantages over the Kyoto agreement. First, enforcement is simplified. Countries can freely choose the extent of their emissions reduction, just as suppliers of ordinary goods and services choose the extent of their contribution to a project like national defense. Second, the sharing of the financial burden is transparent, instead of implicit in the caps.

Two invited talks were given at meetings of the National Research Council’s Committee on Abrupt Climate Change. Further information about the policy program is at the CMI website.



The most difficult challenge in assuring the creativity of a group as diverse as CMI is to develop an integrated research program, rather than a loose collection of separate projects. Weekly research meetings of the entire CMI have been effective in welding together the CMI’s components. These have been exciting, with obvious cross-fertilization. Perhaps the best example is the work on the physics, atmospheric consequences, economics, and policy of leakage from underground reservoirs.

The Carbon Storage Group first characterized the most likely modes of leakage, using their initial models of CO2 in geological reservoirs. As injected supercritical CO2 moves horizontally, some will become sequestered more or less permanently, because it will dissolve or become trapped physically. CO2 that escapes from the reservoir will do so when the moving front encounters an end to the capping formation or a conduit such as a fault or an old well with a failed concrete seal. The material will then migrate upward, spreading horizontally as it moves, until it encounters another relatively impermeable layer, and the whole process will begin again. Thus, a fraction of the material injected into a reservoir will never escape, the remaining fraction will escape with some distribution of time lags, and both the permanently retained fraction and the distribution of time lags will differ from reservoir to reservoir.

The Carbon Science Group has taken this characterization and integrated it into global models of the regulation of future atmospheric CO2. As reported above and at the CMI website, these models show that limits on the rate of leakage are much less restrictive than anticipated in our original proposal. In other words, a reduction of future atmospheric CO2 by 200 or 300 ppm should be possible even with initial average leakage rates between 1% and 10% per year. Not surprisingly, a critical parameter turns out to be the fraction of injected CO2, on average, that becomes chemically or physically trapped. The Carbon Storage Group is working on establishing the value of this fraction. The Capture and Storage Groups, jointly, are extending the work for situations where sulfur is co-sequestered, exploring the implications of larger leaks of H2S or SO2 implied by larger than expected limits on CO2 leaks, and searching for ways to increase the dissolution of CO2, SO2, and H2S in saline aquifers. For the Policy Group, the beneficial effect of heterogeneity is an instance where uncertainty has positive economic value.