The technique of using a "live annulus" for data collection analysis during frac pack sand control completions has evolved quite extensively over the past few years. Field case studies have revealed several anomalies in this data collection and analysis. The following discussion covers more than 100 wells completed in the Gulf of Mexico using mini frac data analysis on both the workstring and "live annulus" data and their relationship to the bottomhole gauge data. The ability to incorporate this "live annulus" data has been facilitated by downhole tool development. The use of "live annulus" data has been viewed by many as being the single most important factor in the successful interpretation of pressure analysis during frac pack operations. Certainly, the elimination of friction in trying to interpret real time data enhanced the ability to make "on-the-fly" decisions. Having a "live annulus" also may help in overcoming some of the water hammer problems in rapid closure situations. However, caution must be used in relying on "live annulus" data. A close look at this summary of data completed in the Gulf of Mexico highlights the need for this caution and questions the degree of it s use. Introduction In the last decade, the evolution of frac packs in high permeability environments have been influenced by lessons learned from data analysis of live surface data and enhanced by bottomhole gauge data. Numerous techniques for analyzing data and designing treatments such as those by Nolte1 and Smith2 have been put into practice for success in frac pack sand control completions. "Live annulus" data analysis has greatly influenced the decision making process for frac pack sand control completions in the Gulf of Mexico. One aspect of "live annulus" data is that during pumping operations, the friction factor for fluids being pumped down a workstring is left out and a more representative profile of bottomhole pressure response is seen less the hydrostatic pressure from the static fluid in the annulus. This is possible from the developments of downhole frac pack packers in the industry that enable the annulus and workstring environments to be in communication with each other simultaneously. As seen in Figure 1, while pumping a Step-Rate-Test, the annulus data is much easier to use for the pressure analysis than the pressure data from the workstring. This is because the dynamically changing friction pressure effects of the fluids in the workstring would be hard to model. Again, the annulus data is simply the bottomhole pressure response less some hydrostatic pressure and as seen in Figure 2, closely follows the recorded bottomhole gauge data. There are instances where data analyzed during mini frac analysis will be performed with "live annulus" data in order to eliminate the early time water hammer effect, that is present with workstring data, hindering the analysis effort in determining closure pressure. Figure 3 is an example of the water hammer effect on workstring data and how it relates to the annulus and bottomhole gauge data. In some situations, closure can happen during this time period due to high leakoff and interpretation of the workstring data can be very difficult. However, caution must be used in deciding that annulus data interpretation will be used under these circumstances. The water hammer effect sometimes can also be seen as illustrated in Figure 4 with "live annulus" data. Figure 5 is an example of how specific tool designs dictate that the annulus data will closely follow the bottomhole gauge data while pumping and when pumping is stopped, the annulus data does not continue to follow the same profile as the bottomhole gauge data. This happens because the specific tool design does not allow for a full flow of pressure in the reverse direction, thus creating a "check valve" or choke affect. Figure 6 is an example of a case where the "live annulus" pressure falloff data did not match that seen by the workstring data or the bottomhole gauge data. Herein lies the dilemma of which data do you use at the time of treatment execution and which is most accurate or meaningful.

Frac-packs, and hydraulic fracturing, have become accepted, successful completion procedures for high permeability formations. To some extent, this success has come despite less than full understanding of the processes. Statements such as "fracture models cannot predict net pressure behavior in soft rocks" are heard. Inconsistencies are blamed on radical departures from "classical" theories of fracturing, and in some instances, this may be warranted. However, it is best to first examine simpler possibilities (Occam's Razor). Radical departures should not be postulated until fracture models routinely address actual geologic/reservoir environments. What is the big difference for high permeability fracturing? Of course, it is not "soft" rock, it is permeability, thus, fluid loss. ALL fracture designs are based on the idea of 1D, i.e., Carter or C/ t, loss, and assume (with no justification) this is valid. High loss is accounted for by high fluid loss coefficients, but using high values for something does not describe the process. One possible cause of the inconsistency might be non-1D, i.e., non-Carter type, loss behavior. Non-1D fluid loss occurs in water injection/water disposal fractures (though "normal" fracture models are still mistakenly utilized in these situations). 1D loss is valid if the fracture propagation is greater than loss velocity, and this condition is NOT true for water flood induced fracturing. Is this true for high permeability fracturing - with fluid efficiency < 10%, even in propped fracturing treatments using viscous fluids? This paper examines this question using a coupled 3D fracture-reservoir model (as described in Appendix A) to accurately simulate fluid loss. We simulate several field cases, review the design/post-analysis based on "traditional" loss behavior, and examine the effect of rigorously simulating loss. The results are used to identify conditions where non-Carter fluid loss is significant, and how to modify designs appropriately. Introduction In what has to be the most quoted "Appendix" in history, Carter1 laid out the basis for "1-D", or "Carter", or "C/Time" fluid loss behavior. This theoretical development has certainly stood the test of time, and has been the mainstay of fracture models and treatment designs for near five decades. Since that work, fracturing has moved from relatively small treatments in moderate-to-high permeability formations (where actually the use of "Carter" fluid loss might have been questionable), to massive fracturing of very low permeability formations (where the "Carter" fluid loss assumption was absolutely justified), to propped fracturing in very high permeability formations for stimulation and sand control (where once again the use of "Carter" fluid loss should be questioned). That is the subject of this presentation, to review the use of this fluid loss assumption by looking at existing work, by studying case histories, and to try to set some limits on when/where "1-D" fluid loss calculations are appropriate.