A satellite-enabled peek at Saudi oil.

Monday, January 28, 2008

Satellite Sleuthing Gone Bad

A recent investors report was produced by Neil McMahon and Ben Dell of Bernstein Research entitled "Bernstein Energy: Saudi and Ghawar from Space! Using proprietary satellite data to test the conspiracy theory". (The report is not available online, but a press release has been widely circulated). It described computer analysis of ASTER satellite imagery taken in 2004 and 2007 and the authors concluded that drilling activity in the Ghawar oil field during the years 2004-2007 was primarily the result of two megaprojects -- specifically the Haradh III and Hawiyah NGL projects -- and not the result of desperate measures to maintain production levels. While data and analysis of this type is certainly welcome, an examination of this report and the stated conclusions has found several shortcomings. Specifically, the assumptions are inaccurate, the data quality is inadequate for the task, the data analysis methodology is suspect, and the conclusions are not supported by the data. But that sounds somewhat negative, so let's look a bit more closely at their claims.

The analysis of McMahon and Dell is based on computer interpretation of "before and after" satellite images and on well drilling statistics from Saudi Aramco. The computer interpretation apparently identified areas of increased well "density" over time as well as areas where the well "density" decreased. This interpretation was performed by digitizing the images and using a computer algorithm to identify features (e.g. mud pit) corresponding to wells. Following this, a density map was generated by determining the number of such features near each point on the map. Shown at right are data from the report showing increases (redder colors) and decreases (bluer colors) in drilling activity in Ghawar in 2007 vs. 2004 (the two years from which "proprietary" satellite imagery was available). The data has been superimposed on the outline of the Ghawar complex with the sub-areas delineated. The red outline indicates the physical area covered in both 2004 and 2007 ASTER satellite image sets.

McMahon and Dell conclude that the increased activity toward the southern end of Ghawar is due to two megaprojects: 1) Haradh III and 2) Hawiyah NGL. There are a few problems with this conclusion, however.

Bad Geography

One big problem is that the red area of higher activity does not encompass any of Haradh. Indeed, the wells of Haradh III are located in the "toe" and are not even covered in the dataset used (red outline). It is unlikely that wells were drilled in Hawiyah and Uthmaniyah to reach oil 50 miles to the south in Haradh.

Bad Eyesight

What about the Hawiyah NGL project? I assume they mean new gas wells which will feed into the NGL plant located east of the existing Hawiyah Gas Plant (which is east of the Hawiyah oil field). These are not technically part of the Hawiyah NGL megaproject, but the Saudi Aramco 2006 Annual Review does state the following:

The Hawiyah gas increment drilling program was initiated, with 32 wells planned.

Unfortunately, it is not possible to distinguish oil wells from gas wells using ASTER imagery (15 meter resolution), although this is possible using higher resolution imagery from the Quickbird satellite (Digital Globe) as I have shown here. Without this differentiation, any conclusions about the type of activity observed (oil vs. gas) are baseless.

Bad Computer Skills

Oil and gas wells are discreet entities. Manually counting wells in the desert is tedious work but provides much more reliable data upon which to form conclusions. Indications that the computer interpretation presented by McMahon and Dell is flawed are found in their identification of broad areas of Ghawar where well density decreased from 2004 to 2007. The authors ascribe this to "dismantling and bulldozing of sites" by Saudi Aramco. However, based on comparisons of high-resolution images with historical Ghawar oil well locations (as shown here), there has been virtually no dismantling of old wells.

Bad Numbers

McMahon and Dell use well drilling statistics from Saudi Aramco to bolster their conclusions. Shown below are the number of wells drilled over the past few years.

From their data, approximately 159 wells of the 329 total for 2006 are assumed to be from Haradh III oil wells and Hawiyah gas wells, along with wells from the AFK megaproject, as shown below:

To obtain this breakdown, McMahon and Dell apparently perused the Saudi Aramco 2006 Annual Review and read the following:

Crude oil increment drilling continued throughout the year, including the completion of 73 wells for the Haradh-III increment, 28 of which were MRCs. Drilling for the AFK increment continued with 54 wells. We began drilling for three other crude oil increments in 2006: Khurais, Shaybah and Nuayyim.

From this, it seems reasonable to conclude that 73 wells of the 2006 total were due to Haradh III. However, the subsequent paragraph adds:

The Haradh-III Increment came on-stream in January 2006, five months ahead of schedule, adding 300,000 bpd of Arabian Light crude oil production capacity to Ghawar, the world's largest oil field, and 140 million scfd of associated gas.

It is unlikely that all 73 wells were drilled and completed within the first month of 2006. Indeed, the following is reported in SPE #105715-MS:

The development was started in February 2005 with the 28 water injection wells at the periphery of the field. Also, 13 dedicated observation wells were drilled and completed with permanent downhole gauges to monitor the fluid front movement in the reservoir. The first producer well was drilled in October 2005.

Also, at a meeting in Bharain in 2004, Saudi Aramco detailed their plans for Haradh III in 2005 as follows:

The year 2005 heralds a critical MRC/smart-well step for Exploration and Production (E&P), with plans moving forward for 14 MRC producers with 14 smart completions, 26 horizontal injectors, and one vertical evaluation well for Haradh Increment-III during 2005. Many MRC/smart-well applications in other Saudi Aramco fields are also on the way.

The onset of water injection occured in November of 2005 according to this report. The field was fully pressurized and tested, indicating that the most of the 28 water injectors and 13 monitoring wells (more than half of the total wells) were in place at that time. Furthermore, early production from Haradh III has been published, and as suggested by the rampup data shown below, it is likely that many (if not most) of the oil producing wells were spudded in 2005 as well.

In Memoriam

To summarize, the Berstein report suffers from a poor choice of data and data analysis method as well as inadequate background investigation into the well counts provided by Saudi Aramco. The questions which remain unanswered are these: what is the extent of oil well drilling in Ghawar and other mature fields for the purpose of mitigating the effects of reservoir depletion (relative to drilling for megaprojects)? Is the amount of drilling activity consistent with mitigating modest depletion rates? The use of high-resolution satellite image analysis with visual identification of wells, as presented here, is a better approach for answering these questions.


Henio said...

"dismantling of old wells." is very rare. In an a field that is still producing no well would be "dismantled" unless the company was short on $$$ or parts.

I do not think this is a problem the KSA has :-)



Shipp said...

Your analysis is convincing and alarming. I looked at some of this area w/ Google Earth and wonder what the black areas around Al Hofuf are. Oil?

JoulesBurn said...


I think that is due to date palms growing on water from the oasis. Try zooming in.


Shipp said...

Dates! More evidence of declining oil reserves. The Saudis are diversifying into the highly profitable date trade.

tom baranes said...

Very interesting article, i wonder if you found some similar idea about Satellite Image Analysis in here http://www.imagesatintl.com/services/satellite-image-analysis/

tom Barnes said...
This comment has been removed by the author.