Tuesday, July 28, 2009

New Refinery On Schedule and Budget

Marathon's refinery in Garyville, Louisiana is undergoing a multi-billion dollar, major project that some refer to as an expansion, but in reality it is the equivalent of building an entire second refinery. The finished project will double the throughput of the current refinery, plus upgrade the processing capabilities.

The remarkable thing is that the $3.2 billion project is being executed on schedule, and with a very modest cost overrun of approximately 25 percent. Such is the experience in an industry that has a learning curve, and actually learns from experience. In fact, no refinery project of this scale has been attempted in the U.S. for more than 30 years. Other new refineries have been built around the world in the intervening period, and some refinery expansion projects have been built in the U.S. I was personally involved many years ago with a major refinery expansion on the Texas coast, which at the time cost approximately $200 million, if memory serves. That project did not expand the crude capacity, but added downstream processing capability.

The same cannot be said for nuclear power plants, of course. The new-generation French nuclear reactor under construction in Finland is a prime example, as it is many years behind schedule and already is experiencing substantial cost overruns measured in the billions of Euros.

The Marathon refinery project was originally expected to cost $2.7 billion in 2006, and that was increased by $500 million to $3.2 billion prior to construction due to escalating costs of materials. There was a world-wide building boom at that time, and costs of cement, steel, and other materials was increasing. The original scheduled completion time was fourth quarter of 2009, just a few weeks from this writing. Recent publications and interviews with the refinery manager indicate that the project is on schedule and will start up later this year. A final cost estimate was reported as $3.4 billion, for a cost over-run of only 7 out of 27, or 25 percent. However, using a more realistic measure, where materials cost escalation are uncontrollable, the cost over-run was 2 out of 32, or approximately 7 percent. Having a major project finish with less than a 10 percent cost overrun is good, and something nuclear power plant builders can only dream of.

In contrast, the last round of nuclear power plants built in the U.S. were many years late in startup, and had cost over-runs measured as multiples of the original estimate. In fact, the notorious South Texas Nuclear Project had a final cost of SIX times the original estimate, $5.5 billion final compared to $900 million initial.

One must wonder what is the reason (or reasons) for one industry having great success at building multi-billion dollar and multi-year projects with a modest cost overrun (25 percent is well within expectations), while another having demonstrated zero ability to control costs (indeed, even to estimate them in the ballpark) or to construct on schedule.

One reason for the nuclear power industry's poor performance given by its apologists are that few nuclear power plants exist, and there is not much experience. But is this a valid reason? There are approximately 436 nuclear power plants operating at this time, with 48 more under construction worldwide. In contrast, there are 676 oil refineries operating worldwide, with a handful under construction. One would think that, after building 436 nuclear plants, some learning would have occurred. But then, apparently not.

One reason could be that nuclear reactor technology changes, and that must be taken into account, as the nuclear apologists are quick to point out. Indeed, the nuclear reactors' design are changing over time. But then, so does the technology for oil refineries. A Hydrocracker has at least six or seven different possible configurations, and Fluid Catalytic Crackers have at least four. Catalytic Reformers also have three or four different designs. Crude Units also have widely different designs, and are almost unique as to each. These are the most expensive process units in a refinery, so these changes matter. These different technologies or designs evolved over the years, just as nuclear reactors evolved. Therefore, changes in design are not the reason.

Perhaps the reason is the regulatory agencies are more strict with nuclear power plants and more lenient with oil refineries. One could make an argument that nuclear power plants have less of a regulatory burden in the area of air and water pollution, as their advocates insist at every opportunity that they do not emit anything into the air except water vapor from the cooling towers, and nothing into the water. Except of course, when they spring a leak and emit radioactive tritium into the groundwater supply. The nuclear apologists do not like to talk about that. Refineries, on the other hand, must obtain air pollution permits for NOx, SOx, and install particulate recovery systems for catalyst particles, to name only a few of the air permits. Water permits are also required for several components in the water streams that exit the refinery. The environmental permitting process is not easy, nor cheap, nor quick for a refinery. Nuclear plants must, however, satisfy the requirements of the Nuclear Regulatory Commission, whose job it is to ensure the plant will not explode, or melt down, or otherwise poison the local and greater community with deadly radioactive particles. Each has its own problems obtaining permits to build, neither are easy. The very near-miss at Three Mile Island had much to do with the NRC's regulations, as that power plant experienced a reactor core meltdown, and very nearly poisoned the Atlantic seaboard. All the while up to that time, nuclear plant designers had insisted that nuclear plants were safe and should be built by the thousands. One shudders to think what would have happened without the vigilance of the NRC.

Perhaps, as some suggest, the issue is that nuclear power plants cost so very much more than does a refinery, and this leads to problems associated with mega-projects. As the Marathon refinery project shows, however, the project average expenditures was $1.7 billion per year ($3.4 over two years). The latest cost and schedule estimate for the South Texas Nuclear Project expansion is a cost of $13 billion and completion in only six years. Leaving aside the fact that both of those figures are wildly optimistic, and the final cost will be on the order of $25 billion and the duration will be 10 to 12 years, the result is just over $2 billion per year ($13 over six years). There is very little difference between $1.7 billion and $2 billion per year in a major project. Thus, it cannot be the money spent per year as a factor.

What it really comes down to, and properly in my view, is that refineries provide a much-needed and clean-burning suite of products, from propane, gasoline, jet fuel, diesel fuel, home heating oil, lubricating oil, greases, waxes, and petrochemical feedstocks that do not get burned but are converted instead into highly valuable chemicals that produce an incredible number of essential products in the modern world. Plastics and pharmaceuticals are just two of the thousands of products. Refineries are viewed by the courts as ordinarily hazardous, but nuclear power plants are viewed as ultrahazardous. (I plan to have much more to write on this distinction very soon). This results in more lawsuits from anti-nuclear groups, however, refineries also experience lawsuits for their construction projects from anti-refinery groups.

Also, for whatever reason, it appears that nuclear power plant construction teams cut corners with the result that the work is shoddy and must be torn out and rebuilt according to appropriate codes and design specifications. One wonders why the construction teams do not simply do the job correctly the first time, demonstrate that their work meets the inspectors' requirements, and move on smoothly to the next item on the schedule. This is how refineries are built, and this is no secret.

The ultimate test is building a modern, new-generation nuclear power plant in the U.S. under the auspices of the NRC, and not compare the activities overseas such as South Korea, China, or France. One would expect that, with 48 nuclear plants under construction world-wide at this time, and some only recently started up, that the cost estimating and construction scheduling would be honed to a fine science. All the rest is just talk.

The first such nuclear power plant in the U.S. may actually be the expansion of South Texas Nuclear Project, which is planned to have reactors 3 and 4 added to the existing 1 and 2. The project proponents claim they can build the expansion for $13 billion, as stated above, which I do not in any way believe. The actual cost will be approximately $25 billion, and its owners will have serious problems selling the power from the plant. The reactors are to be 1,350 MW each in size, for a total of 2,700 MW. Per solid and reliable cost estimating methodology presented by Craig A. Severance, CPA, the STNP expansion will cost $10,000 per kW, or $27 billion. Of course, as the project experiences delays in construction, interest costs will mount and the cost could easily exceed $30 billion. Also, Mr. Severance's cost estimate did not include (because the new requirement was not yet law) the design to withstand the impact from a large commercial aircraft, for the reactor dome, the cooling system, and spent fuel storage areas. As I wrote elsewhere on this blog, that is a tall order (and very expensive) if the cooling towers must withstand a large aircraft impact. By large, the regulation refers to a fully loaded Boeing 747 or an Airbus 380.

The people of San Antonio, who are supposed to own a large portion of the STNP expansion, will pay very high power costs, perhaps the highest in the nation, if the expansion is built. This does provide a silver lining, as the high power costs will give great incentives for electric customers to install self-generation systems, such as cogeneration, combined heat and power systems, solar panels, and wind-generators where appropriate.

It is also quite instructive that the STNP is located only a few miles from Corpus Christi, and that offshore Corpus Christi for 50 miles out into the Gulf, and 75 miles south, lies one of the U.S.' greatest and consistent wind resources. The wind there averages a Class 5, or 20 miles per hour. The U.S. MMS is preparing to lease that offshore area to wind-power developers.

It is also quite instructive that the leading competitor to nuclear power, natural gas-fired plants, are having a banner time due to the very low prices of natural gas. The current economic recession is only a small part of that, what is a major reason for low gas prices is the discovery and production of huge amounts of gas from shale formations, and the LNG plants that are now online and producing around the world. Natural gas is easily capable of base-load power generation, and more importantly, in load-following.

These are interesting times, and if indeed the STNP expansion obtains its financing, and construction permits, it will be fascinating to see who remains as a customer 10 years later. And, when STNP fails, other utilities around the nation will take note. This could be the last nuclear plant built in the U.S., ever.

UPDATE 1, August 3, 2009: Today Marathon announced the project will cost $3.7 billion and is 91 percent complete with startup scheduled as planned for 4Q 2009. Therefore, cost overrun is 10 out of 27, or 37 percent. The additional costs ($300 million) are for materials price escalation. Link is here.

UPDATE 2, October 30, 2009: The Marathon Garyville refinery is in the startup phase, indicating on-schedule completion. Good job, guys. Now, if only nuclear power plant builders (a far simpler, less complex process) could learn a few lessons from the oil refining industry. Apparently not.

At latest press, STNP expansion is now projected (by Toshiba, the reactor vendor) to cost $17 billion - not $13 billion as previously stated by them - representing almost a 33 percent increase. This expansion project has yet to turn one shovel of dirt. One would think that Toshiba would know the cost, as their technology is loudly touted as "proven." Again, apparently not.

Roger E. Sowell, Esq.
Energy and Climate Change Attorney


M. Simon said...

One thing to keep in mind re: nukes is that the reactor and its associated control system costs about 20% of the total. The rest is BOP. Steam generator, turbine, condenser, main feed pump, etc.

BTW extensive air quality monitoring is required for nukes. Gases, and particulates.

As to why they cost so much? I couldn't say. What I learned about nukes was from being a Reactor Operator in the US Navy and I have not kept up much since then ('66 time frame).

My guess is there is no one like Rickover with his attention to detail in the business any longer. Even in a process with a lot of regulatory approvals required it ought to be possible to keep things moving by PERT charting the process and applying resources to milestones that are in trouble.

One other possibility is that they are handing out contracts to the lowest bidder and not to the most qualified bidder. i.e it is worth paying 10% more to get the job done right the first time.

You might want to visit Idaho Samizdat: Nuke Notes and ask him some questions.

Roger Sowell said...

Mr. Simon, I stay abreast of developments with STNP because I am from that part of Texas, have family there, have many friends there, and will likely retire there one day.

There is a website (how good, I cannot say, but see below) that has a brief history of that plant; here is an excerpt:

STNP began construction in 1976, with a promised completion in 1981. Instead, Unit 1 construction was finished seven years late in 1988, and Unit 2 in 1989. The initial cost estimate made in Dec. 1973 was $964 million-but actual costs ballooned more than five-fold to $5.5 billion."

source is http://www.nukefreetexas.org/stp_record.html

This was one of the last of the 104 plants to be built, and obviously benefited not the slightest from any lessons learned from earlier efforts. A six-fold cost overrun is beyond ridiculous.