The potential for tremendous economic impact has fueled research to improve safe production rates to control the risk of sand control failure. According to Dave Tiffin PhD, a senior member of the BP Gulf of Mexico shelf sand control and completions team in the Upstream Technology Group, along with associates Michael Stein and Xiuli Wang, using data from over 200 sand control wells has allowed them to develop a sound technique to determine safe production rates.
In a presentation to the SPE reservoir study group recently, Tiffin said that although working with field data was extremely difficult, their analysis led to development of a simple function of volumetric fluid flow per unit area of screen in units of velocity (flux). The function proved reliable at separating wells operating safely from those resulting in damaged screens or unacceptable amounts of produced sand. He noted that this is a work in progress that is continuing to improve.
“What we found when we started this work was a wide variety of different sand control guidelines in BP, as well as other operators,” Tiffin says. “Most all (operators) used pressure-based guidelines, and if you asked them what the bases were, we heard anywhere from 200 psi to 1,500 psi. The only thing consistent was inconsistency based on experiences and rules of thumb.” These pressure-based draw-down limits are either ineffective for managing risk of well integrity or they unnecessarily constrain well productivity.
A lot of new sand control developments use the same guidelines to design facilities and tubing sizes. Reservoir engineers ask the completions engineer, “How are you going to design the completion? What kind of skin are you going to have? What kind of drawdown will you have?” The reservoir engineer plugs this information in the reservoir model and often designs everything, including tubulars and facility capacity, around this conversation. Is this the best way to do it?
In acquiring the data, BP concentrated on the cased-hole sand control completion with frac packs and gravel packs. The company did not cover open-hole completions at this time.
“First, we looked at good quality completions with evidence that the annular space was gravel filled and the screen assembly got to the bottom with little or no damage,” Tiffin says. “Just those two things alone seemed to eliminate a lot of discrepancies. We tried to keep it as simple as we could and did not include skin in our initial analysis, due to unreliable pressure transient data and its interpretations.”
Correlating data
The first attempt at correlating data tested how well the data lined up with drawdown. After plotting, it was clear that drawdown applied in this way does not help predict safe operating conditions, nor can it be used to optimize production rates, Tiffin says.
“The team concluded that although drawdown is not a good parameter to predict whether a sand control completion will fail, drawdown or pressure drop across the completion is a key parameter in determining when the sand matrix fails and the individual sand grains can be transported by the fluid flow entering a well. This may be the basis of using drawdown to control wells with a sand control completion,” Tiffin says.
A plot of drawdown versus perforation velocity incorporating these two adjustments confirms screen erosion failures and wells with restricted production because sand production increases quite dramatically as perforation velocity increases.
Many models and predictive techniques make this determination based on rock strength measurements. Depletion forces also weaken the sand matrix, resulting in a well capable of producing sand free at high drawdowns early in well life, but failing later in life after the reservoir is partially depleted with the same or smaller drawdown. Also, just because a rock fails does not mean sand will be produced.
Sand control completions, like frac packs and gravel packs, are designed to contain the sand whether the reservoir has failed or not. For this reason, using drawdown to control wells with effective sand control completions only makes theoretical sense when there is either no sand control completion or an ineffective one in place, or in a very compressible, highly depleted formation, where wellbore and screen collapse is a risk.
With a good quality completion in place, the analysis of screen failures indicated that screen erosion was by far the most common screen failure mechanism. Fluid that flowed through the screen with a small amount of fine sand particles greatly accelerated the screen erosion.
Erosion of the metal screen by solids is primarily a function of the kinetic energy of the downhole fluid. For a given fluid, kinetic energy increases with the velocity or flux of the fluid.
“Flux-based production limits were successfully applied to sand control wells in BP by Fraser Elliott and a team of engineers at one of our offshore fields,” Tiffin says. This team used experimental results from Southwest Research Institute to obtain a maximum flux rate through the screen at which no damage would be expected for a 10-year life.”
Drawdown limits were changed from 750 psi to the maximum calculated flux at the screen basepipe diameter. BP calculated the inflow area of the screen at the base pipe and then corrected it using skin calculations to estimate partial penetration and percent wellbore area flowing. BP increased production results by applying these new flux-based limits, which were quite successful for the original completions in this thick fairly homogeneous sand.
BP decided to develop guidelines for better operation of sand control wells. Tiffin says BP collected additional historic field data to determine what key variables controlled success and failure of screen life in the field.
Most of the data from the GoM and Trinidad included all sand control completion types. However, success and failure proved difficult to define, so wells were grouped into the following categories:
• Poor completion quality
• Likely erosion failures
• Wells with rates restricted due to sand production
• Producing small amounts of fine sand, but no need to restrict rates
• No sand, no problems.
“Since flux is volumetric rate per unit area of screen, determining downhole rate from surface rate was relatively easy with the use of accurate formation volume factors. All flowing phases were considered in this volumetric calculation, but determining what area of screen to use was not so straightforward,” Tiffin says.
BP took two important steps to adjust the inflow area of cased hole sand control completions before success and failure successfully correlated with flux, Tiffin explains. The first adjustment calculated perforation area rather than basepipe inflow area. This adjustment resulted in more realistic values of flux at which one might expect erosion damage.
The second adjustment to inflow area not considered earlier proved critical to success. Observers recognized during analysis of failed wells that failures were occurring in wells with high permeability streaks. They also observed while studying a failure in the previously cited offshore development where BP first used flux-based limits that existing operating limits, though quite successful for big blocky sands, did not work in heterogeneous sand. This led to the conclusion that inflow area along the wellbore is a function of kh and varies with reservoir heterogeneity.
“We attempted to account for this with the introduction of heterogeneity index, Tiffin says. This correction led to the correlation of wells that failed with high-calculated flux values.
For most of the data, this heterogeneity index came from the same data used to develop reservoir models and included log responses from the porosity, resistivity, and gamma ray logs. When possible, log response tied to actual core data. An alternate method to determine the heterogeneity index is to use production log test data. Production log test data correlated well with heterogeneity calculations, but data was limited, especially in high rate wells.
In addition to this heterogeneity index, BP considered a fixed percent of perforations open and flowing and only considered net pay to contribute. The company uniformly applied open perforation assumption to all of the cased-hole completions in the database.
A plot of drawdown versus perforation velocity incorporates these two adjustments, which confirms screen erosion failures and wells with production restricted because of sand production increases quite dramatically as perforation velocity increases. The company did not observe any difference in calculated flux rates required for failure between frac packs and cased-hole gravel packs.
For gas wells, the onset of a high failure rate appears to occur around 20 ft/sec. This value is roughly 10 ft/sec for oil wells, which correlates the onset of observed erosion damage in premium screens during testing at Southwest Research.
C-factor correlation
BP observed a difference between gas and oil wells. Erosion experts have recognized this and developed a C-factor to account for differences in erosion with different fluid phases.
C = (density mix) 1/2 x (maximum perforation velocity), where density is in lb/cu ft and perforation velocity is in ft/sec.
Errosion experts obtain mixture density by weighting each phase density by its volume percent of flow. The resulting mathematical term is the square root of the kinetic energy of the fluids that carry sand into the screen.
It is important to keep in mind that all data correlated sits within a set of standard completion parameters. Perforation density, screen diameter, annulus pack thickness, and fluid viscosity are a few examples of data that generally fit within a well-defined range. It would not be a good idea to use this correlation on wells with parameters outside those used to build the correlation.
Production rate guidelines assume a failure mechanism based on screen failure due to too high a flux through the screen, resulting in erosion. BP identified other failure mechanisms during the investigation of field data.
A series of wells failed in one GoM field because of massive reservoir depletion and compaction forces. Wells failed within a short time in an entire block as a result. Casing collapse, screen collapse, and other problems in well construction not designed to withstand these tremendous forces caused well failure. The failure was not solely a result of pulling the wells too hard, but too much depletion given overburden forces. Little could have been done short of being aware of this inevitable collapse and either shutting down production or injecting a fluid to keep reservoir pressures sufficiently high to prevent collapse. This was the only documented case of this type of failure that found, though there were vague references to similar failures. As a result, BP suspected that this failure mechanism is rare, especially in cased hole completions.
Screen failure by erosion is only one of countless failure mechanisms. Others include: corrosion, QA/QC manufacturing defects, installation damage, plugging by scale, asphaltenes or paraffin, and countless others. All these need to be considered during the process of establishing production limits in sand control wells.
Editor’s note: Additional information can be obtain in SPE 84495.
