Laboratory testing of fluid leakoff enables optimum recommendations for fracturing fluid composition in addition to accurate predictions of relationships between leakoff rates and formation and fracturing-fluid properties. Static fluid-loss measurements, the present standardized testing method, provide inadequate results for comparing fracturing-fluid materials or for understanding the complex mechanisms of viscous fluid invasion, filter-cake formation, and filter-cake erosion. Previous dynamic fluid-loss studies have inadequately addressed the development of proper laboratory methods, which has led to erroneous and conflicting results. This paper (Part 1 in a series) discusses the results of a consortium (1989 to 1994) that was established to help us understand and allow modeling of the fluid-leakoff process in hydraulic-fracturing applications. Part 1 summarizes over 300 laboratory experiments structured to study the effects of cell design, fluid-preconditioning (both high-shear and low-shear regimes), and core thickness on fluid leakoff. The results help define appropriate laboratory equipment and testing procedures for realistic modeling of the filtration process of fracturing fluids.
The rate of fluid leakoff to the formation during a hydraulic fracturing treatment is one of the most critical factors affecting fracture geometry and treatment performance. The filtration rate during a fracturing treatment affects the designed treatment size, optimal proppant schedule, and resulting proppant distribution within the fracture. Excessive fluid loss may result in insufficient fracture geometry, while low leakoff may result in poor proppant distribution within the fracture as a result of long fracture-closure times. For estimates of fluid leakoff rates, field measurements (commonly known as "minifracs") are required before each main treatment to determine estimates of average filtration rates. Accurate analysis of the minifrac data requires a good understanding of the filtration mechanisms only available through laboratory testing, since leakoff may differ significantly with injection volume. Additionally, the ability to use the minifrac data to estimate filtration rates of other fluids or additives and other well conditions (permeability, temperature, and pressure) is presently not feasible, since these variables may dramatically alter the filtration rates. These difficulties require an understanding of filtration mechanisms that can only be determined from laboratory experimentation.
Numerous studies have been performed that attempt to develop acceptable laboratory fluid-leakoff databases; however, significant laboratory variations from poor experimental equipment design and procedures have made developing the required relationships to field conditions unattainable. This paper (Part 1) describes extensive laboratory testing for the development of a reproducible and accurate testing procedure. This paper also addresses classical filtration theory as applied in most fracturing simulators, limitations to the classical theory, previous experimental studies, laboratory equipment and procedures, discussion of results, and conclusions. Discussion of results will be separated into sections on cell effects, fluid preconditioning effects, core thickness effects, and recommendations for a standardized fluid-leakoff testing procedure. Later papers in this series will cover the results of shear rate, permeability, pressure, fluid composition, temperature, fluid-loss additives, and fluid-loss modeling. The first three variables are described in detail in SPE 36493, Part 2 of this series of papers.
In the industry fracturing-fluid leakoff is normally modeled with three filtration-resistance coefficients (known as fluid-loss coefficients):
the resistance to fluid loss from the filter cake,
the resistance to flow in the invasion zone, and
compressibility of the noninvaded zone (Fig. 1, Page 12).
The noninvaded and invasion-zone coefficients are normally calculated with the expressions in Eqs. 1 and 2 (Page 2). The invasion-zone expression assumes Newtonian-filtrate properties.
P. 805
