Dynamic fluid-loss measurements improve frac design

Sanjay Vitthal, Patrick Walker
Halliburton Energy Services
Duncan, Okla.

Data from dynamic fluid-loss testing improves fracture treatment design. These data represent more accurately the true fluid loss in a fracture and experiments show that the relative benefits of various fluid-loss additives are different under dynamic conditions.

Four principal factors controlling fluid loss are:

1. Fracturing fluid type

2. Rock permeability

3. Fracture-fluid shear rate

4. Pressure drop driving fluid loss.

The relative effect of shear rate depends on the fracturing fluid (linear or crosslinked) and its viscosity. Generally with greater viscosity, the effect of shear rate on fluid loss is greater. Increasing shear rate tends to increase fluid loss.

Dynamic fluid-loss data have improved the agreement between expected fluid-loss and field-measured fluid loss.

Fig. 1 [49277 bytes] shows fluid-loss testing equipment for dynamic flow conditions. The dynamic fluid-loss loop addresses the effects of fluid preconditioning and shear rate on fluid loss and simulates the shear and temperature history of the fracturing fluid as it goes down the tubing and into the fracture.

Test procedure

Fracturing fluid is pumped through 1/4 and 1/2-in. stainless steel tubing before it enters the fluid-loss cells. The 1/4-in. tubing is at ambient temperature and simulates the high-shear environment in the well bore. Various lengths of 1/4-in. tubing can be selected to simulate the effects of different well bore lengths.

The 1/2-in. tubing is immersed in a heating bath, where it simulates the fluid flow in the fracture near the well bore. The Newtonian shear rate in this 1/2-in. tubing is identical to that of the fluid-loss cells.

A typical residence time in the heating bath is 15 min. This fluid preconditioning system is believed to give reasonable modeling of early time fluid properties (near-well bore fluid loss). For shear-sensitive fluids, the length of the high shear tubing can dramatically affect fluid-loss results.

The individual fluid-loss cells contain 1.5-in. diameter cylindrical core samples with a standard core length of about 1.0 in. Six cells are run concurrently and are plumbed in series so that all the fluid passes across each core.

Each cell has a smooth transition piece to eliminate fluid entrance effects. The smooth transition pieces help ensure that the fluid maintains a constant shear rate while undergoing the transition from pipe flow to the slot flow in each cell, which results in even fluid flow across the entire core surface.

Previous testing has shown that the absence of flow transition pieces adversely affects fluid-loss results. The sudden change from cylindrical flow to slot flow can result in fluid "jetting" across the core face at unpredictable shear rates. This "jetting" distorts results and creates a visibly uneven filter cake.

Fluid-loss filtrate from each core passes through a small heat exchanger and a back pressure regulator before being collected on an electronic balance. Heat exchangers and back pressure regulators are necessary for high-temperature testing.

A computer data acquisition system continuously collects the electronic balance data and stores the pressure, temperature, flow rate, and pipe viscometer data. Filtrate mass, fluid flow rate, and cell pressures are stored as often as every second.

Fluid-loss stages

Fluid loss typically occurs in two distinct stages. The first stage corresponds to the fracturing fluid building up a concentrated layer of polymer and fluid-loss additives, or filter cake, on the rock surface. Because the rate of fluid loss is relatively high during this stage, it is commonly referred to as "spurt loss."

The spurt volume (Vspt) indicates how much fluid loss is required to build up a competent filter cake. Formations with either high permeability or natural fractures commonly exhibit high spurt loss.

In the second stage, fluid loss is controlled by the established filter cake. The filter cake resists additional fluid loss because its permeability is much lower than formation permeability. The wall-building coefficient (C_w) indicates the fluid-loss rate through the established filter cake.

Static test

Because static tests are simpler to run, most past fluid-loss testing has been static. In static fluid-loss tests, a core is placed inside a cell (Fig. 2 [22110 bytes] ), which is then filled with a fracturing fluid. After the cell is heated and pressurized, a valve is opened on the back side of the core to initiate fluid loss. No shearing of the fracturing fluid occurs during the test.

Fluid passing through the core is carefully collected and measured with time, often resulting in a curve similar to Fig. 3a [44871 bytes] .

Static fluid-loss tests have led to the development of fluid-loss models for almost all fracture simulators in the industry. The models typically show that fluid loss is controlled by the following three different mechanisms:

1. Resistance to fluid loss caused by the filter cake permeability

2. Resistance offered by fracturing fluid filtrate invasion into a reservoir

3. Resistance to fluid loss caused by fluid injection into a compressible reservoir.

These fluid-loss models further demonstrate that fluid loss from each of these mechanisms is proportional to the square root of time. Therefore, a fluid-loss coefficient can be defined for each mechanisms as follows:

• C_w-Contribution of the filter-cake effect on fluid loss

• C_v-Contribution caused by the formation's resistance to fluid invasion

• C_c-Contribution caused by resistance to injection into a compressible reservoir.

These individual coefficients are then combined to obtain the overall fluid-loss coefficient (C_eff).1

Dynamic fluid loss

Under static conditions, the fracturing fluid is stationary, and the filter cake can grow indefinitely. During a fracturing treatment, however, the fluid moves over the face of the filter cake as it flows through the fracture.

Fig. 3b [44871 bytes] shows the effect of shear rate on fluid loss for a 40 lb/1,000 gal borate-crosslinked hydroxypropyl (HPG) gel. The data indicate that as shear rate increases, fluid loss increases dramatically. The shear rate of the flowing fluid creates a tangential force on the filter cake.

As the filter cake grows on the rock face, it reduces the channel width through which the fluid is flowing. This reduced width increases effective shear rate over the filter cake, which in turn, increases shear forces on the filter cake.

As shear rate increases, the imposed tangential force increases the erosional forces on the filter cake and, therefore, opposes additional filter cake growth. The shear rate also tends to create a "lift" force on the polymer "particles," which will also prevent them from depositing. These phenomena cause a thinner filter cake and higher fluid loss.

Dynamic fluid-loss experiments in the laboratory support the concept described previously. Testing indicates that under dynamic conditions, the filter cake initially builds up similar to a static filter cake. However, as the shear forces on the filter cake increase, the filter cake growth slows or stops, leading to higher overall fluid loss.

As filter cake growth slows, fluid loss gradually changes from being a linear function with the square root of time to a linear function with time. Such a change is not observed in static fluid-loss tests.

Table 1 [22196 bytes] compares the fluid-loss coefficient (C_w) of a borate-crosslinked HPG on a low-permeability, less than 1 md, sandstone measured under static and dynamic conditions. Table 2 [6966 bytes] tracks the effect of shear rate on the spurt and C_w of high-permeability sandstone.

The spurt loss under dynamic conditions is slightly higher than that measured under static conditions. In this case, the dynamic C_w is an equivalent number that is defined as the fluid-loss coefficient that matches the total fluid loss during a 60-min test.

Dynamic fluid-loss tests are more complicated and more realistic than static fluid-loss tests because the fracturing fluid is subject to appropriate shear and temperature preconditioning, and the dynamic fracturing fluid flows across the core surface.

Preconditioning the fluid is important because it simulates the process a fracturing fluid undergoes in an actual treatment.

Fluid properties directly affect fluid loss. Flow across the filter cake surface while fluid loss occurs simulates conditions in a downhole fracture. Fluids in static tests generally either have no preconditioning or are allowed to remain static during cell heating. Both of these factors affect fluid loss.

Applications

Fluid testing under dynamic conditions can lead to more effective and economical selection of fluids and fluid-loss additives. In addition, the fluid-loss coefficients, as measured under dynamic conditions, agree closer with those determined in the field through minifrac data analysis.

Dynamic fluid-loss testing becomes useful by incorporating both fluid preconditioning and dynamic flow conditions. As seen in Fig. 3b [44871 bytes] , fluid-loss relates directly to the shear rate in a fracture. However, shear rate in the fracture varies with time and distance from the well bore. Shear rate is also a function of the pumping rate and the formation being fractured.

If the shear rates are high over a significant portion of the fracture, dynamic flow conditions should affect fluid loss and possibly alter the geometry of the fracture.

Fig. 4 [35713 bytes] profiles the shear-rate in a fracture at different times during a treatment. Note that near the beginning of the treatment (25% pumped), the shear rate is relatively high. The portion of the fracture where spurt loss is occurring also has the highest shear rates.

Throughout this treatment, the shear rate is high enough (greater than 50 sec^-1) over a large enough portion of the fracture that dynamic fluid-loss results should be used. In cases where the shear rate is low (less than 15 sec ^-1), the dynamic fluid-loss data should still be used because of fluid preconditioning.

Minifrac examples

In the first example, a minifrac/fluid efficiency test was run to determine fluid-loss coefficients in a low-permeability gas formation. Table 3 [9157 bytes] lists the well conditions. The treatment involved a 40 lb/1,000 gal gel of zirconate-crosslinked carboxymethylhydroxy propyl guar (CMHPG).

The treatment was pumped down tubing. The annulus was kept open to estimate the bottom hole pressure. The square root of time and G-function plot analysis indicated that fracture closure occurred at about 4,200 psi at surface. The closure time, pressure data, and other formation properties were incorporated into a fracture simulator to calculate the fluid-loss coefficient.

Fig. 5 [36465 bytes] plots the net pressure match. The calculated filter cake coefficient (C_w) was 0.0055 ft/min^0.5 which agrees reasonably well with the dynamic fluid-loss data in Fig. 6 [33504 bytes] .

In another case, a minifrac test was conducted in a high-permeability, oil-producing formation before a frac-pack treatment (Table 3). Treatment fluid was a 30 lb/1,000 gal HPG with borate crosslinker.

A plot of the pressure decline-vs.-square root of time indicated that fracture closure occurred at about 3,630 psi. The net pressure response during the minifrac was modeled with a fracture simulator (Fig. 7 [36633 bytes]). The match indicated that the pressure response was best matched with a 0.28 gal/sq ft spurt loss and a 0.004 ft/min^0.5 C_w.

This match again agrees reasonably well with the dynamic fluid-loss data. In contrast, a test with the fluid under static conditions indicated a 0.0027 fpm^0.5 C_w.

The Authors