Technical Report NTB 87-03
Design of a single borehole hydraulic test programme allowing for interpretation-based error
Hydraulic testing using packers in single boreholes is one of the most important sources of data to safety assessment modelling in connection with the disposal of radioactive waste. It is also one of the most time-consuming and expensive. It is important that the results are as reliable as possible and as accurate as necessary for the use that is made of them. There are many causes of possible error and inaccuracy ranging from poor field practice to inappropriate interpretation procedure. The report examines and attempts to quantify the size of error arising from the accidental use of an inappropriate or inadequate interpretation procedure. In doing so, it can be seen which interpretation procedure or combination of procedures results in least error. Lastly, the report attempts to use the previous conclusions from interpretation to propose forms of field test procedure where interpretation-based errors will be minimised.
Hydraulic tests (sometimes known as packer tests) come in three basic forms: slug/pulse, constant flow and constant head. They have different characteristics, some measuring a variable volume of rock (dependent on hydraulic conductivity) and some having a variable duration (dependent on hydraulic conductivity). A combination of different tests in the same interval is seen as desirable. For the purposes of assessing interpretation-based errors, slug and pulse tests are considered together as are constant flow and constant head tests. The same method is used in each case to assess errors. The method assumes that the simplest analysis procedure (cylindrical flow in homogeneous isotropic porous rock) will be used on each set of field data. The error is assessed by calculating synthetic data for alternative configurations (e.g. fissured rock, anisotropic rock, inhomogeneous rock – i.e. skin – etc.) and then analysing this data using the simplest analysis procedure. In this way the error can be quantified as the derived value compared to the value used to calculate the synthetic data. The principal value of interest is that of hydraulic conductivity (K) with specific storage (Ss) of minor concern except as the basis of a coherent understanding of the test. A further benefit of this approach is that synthetic data can be assessed as to whether the particular configuration under examination will yield data curves which are distinctive.
The most likely causes of error in the analysis of slug/pulse tests are considered to be skin effect, non-cylindrical flow (including packer bypass) and fissured rock. It is concluded that skin effect will strongly affect the derived value of K either where the effective radius (re = the radius of an imaginary incompressible tube in which an open water level fluctuates during the test) is small (i.e. less than 0.5 mm), the skin is thicker than 10 mm or the ratio of the K of the rock to the K of the skin is greater than 100. It was also seen that skin effects only resulted in distinctive results when the responses were "transitional" between a skin-only and a rock-only response. Non-cylindrical flow will only result in significant errors where the radius of the borehole times the component of K parallel to the borehole is greater than the length of the tested zone times the component of K perpendicular to the borehole. Fissured rock results in maximum errors in derived K of two times, the derived value always being an overestimate. The data curves from fissured rock are not distinctively shaped but yield characteristically low values of specific storage.
The principal imaginable sources of error in single boreholes constant flow/head tests are skin, wellbore storage, fissuring and non-cylindrical flow. The error resulting from skin depends on the method of analysis. Pseudo steady state methods such as those of Hvorslev (1951) or Moye (1967) are the most prone to error whereas methods which use the rate of change of head/flow are theoretically reliable. They are theoretically reliable only so long as the period of wellbore storage has been correctly identified. Fissuring and non-cylindrical flow have already been examined by other authors (see Braester and Thunvik, 1984) and found to yield insignificant errors.
The question of whether constant flow or constant head tests are preferable is examined. It is concluded that the most reliable form of test analysis and interpretation is based on a test combining drawdown and recovery. This is best achieved using the constant flow form of the test but carried out at the same approximate head.
The constraints on the duration of both forms of test (slug and constant flow) is examined and it is concluded that the duration should depend on the transmissivity (rock K times zone length) of the tested zone. The control of test duration is based on a predetermined amount of pressure recovery after a period of pressure disturbance. The report covers other aspects of test design, such as testing in patchy rocks, the use of the "gradient method" to analyse slug/pulse tests and the interaction between form of the required data set and field practice. The report recommends a form of field testing which consists of a pulse test followed by a constant flow test which includes a period of recovery. This form is thought to provide the best opportunities for discerning possible causes of error and allowing reasonably error-free analysis under the conditions proposed.