Goran Tuomas
Division of Renewable Energy, Lulea University of Technology, SE-97187 Lulea, Sweden
Received 28 November 2002; received in revised form 10 August 2003; accepted 18 August 2003

ABSTRACT

Drilling with water driven down-the-hole (DTH) hammers is a recently developed method for competitive production of boreholes. In order to prevent large amounts of water being used during operation, the drilling fluid is here directly processed into a quality acceptable for reuse. The effectiveness is evaluated in well drilling with a mobile prototype water cleaning and pressurising unit. Especially the presence of abrasive particles in the fluid can drastically reduce tool life and make the method inefficient. The vital significance of this relation has called for detailed studies and a process simulation model for determining particle concentration and size distribution has been developed. This paper describes the model and how it is applied. Simulation results of different system configurations are also presented. © 2003 Elsevier Ltd. All rights reserved.
Keywords: Drilling; DTH; Hammer; Down-the-hole; Particle; Flow; Water; Simulation

1 Introduction

The technique of using water instead of air as an energy carrier to DTH-hammer tools has been known for years. However, technical difficulties associated with corrosion, cavitation and wear have made it difficult and/or costly to put these ideas into practice. This situation began to change in the early 1990s when the Swedish mining company LKAB started to use water driven DTH-hammers for production drilling of blast-holes. The use of the hammer-tool also meant continuous evaluation and improvements of the system, which today is a highly cost-effective and competitive drilling method. Today, more than 5-million meters of blast-holes have been drilled with the water driven hammer tool within the Swedish mining industry.
There are many advantages with this method; the most important are its cost-effectiveness and competitive performance. The technique offers high penetration rates and low energy consumption as well as the possibility to drill to virtually any depth (Tuomas and Nordell, 2000). The working environment is improved since dust is eliminated and the air is free from oil residues. However, one disadvantage is that a large flow rate of preferably high quality water is required to drive the hammer tool. For instance, an ordinary 4-inch hammer-tool requires between 0.2 and 0.4 m3/min to achieve a competitive rate of penetration. This means that the water should be recycled when this drilling method is used in locations with limited water access and/or when waste disposal is difficult to accomplish (Fig. 1).

Fig. 1. Principle flow in a drilling system with re-circulation

The concentration of particles in the drilling water depends mainly on the actual water flow rate, penetration rate, and the density of the drilled rock. Mass concentrations (w/w) between 4 and 12% are common for rock drilling with an ordinary 4-inch hammer. This corresponds to approximately 13-27 kg/min particle flow, which means that high-capacity cleaning equipment has to be used. The particle size distribution varies with a certain number of factors. Rock characteristics, drill bit design and impact energy, are some of them. An important limiting factor during vertical or inclined drilling is the speed of the flushing water, since this must be larger than the particles settling speed. Otherwise the particles will settle in the borehole and will be re-crushed by the drill bit until the size is small enough to follow the flow. Particles generated during typical 4-inch well drilling are usually smaller than 1 mm with mass median sizes (dso) at approximately 0.1 mm.
For the technique to be successful, the fluid cleaning system must be correctly designed and implemented since fluid quality directly affects component life. Abrasive particles and/or aggressive chemical substances in the feed water significantly reduce tool life, especially when ordinary tools made of hardened steel are used. It is, however, possible to use tungsten carbide as tool material, but this increases the cost and this material is, therefore, normally only used in mud driven tools. For this reason, knowledge of how different water related parameters affect the life of a given tool or material is of vital importance when designing cost effective systems.
Interesting data have been obtained from practical use of these tools, especially within the mining industry where automated drill-rigs produced millions of meters of 4-inch blast holes. Results from water-analysis and data of the corresponding tool-life, show that time between repairs corresponds to approximately 1500 drilling meters in hard rock when the feed water contains maximum 0.02% w/w solids. The mean penetration rate during these drillings was 0.9 m/min, which gives a total of approximately 6 million piston blows between repairs, since the piston blow frequency is about 60 Hz. Other experiments have shown that the life was drastically reduced by large amounts of solids in the feed water. For example, life less than 100 drill-meters have been measured when the feed water contained about 0.5% w/w solids (Oderyd 2001).
To evaluate the possibilities of this system, a complete mobile prototype service unit for use with low-cost clear-water hammers has been constructed (Tuomas 2001). The unit includes all components required for efficient drilling, i.e. systems for both pressurising drilling fluid and particle-fluid separation (by a lamella thickener and a hydro-cyclone unit) to enable recycling. The prototype unit is presently undergoing initial operational tests in order to establish the relation between tool life and particle content in the drilling fluid. System characteristics for the prototype were estimated by simulations with a process model, implemented within the Matlab Simulink math package. Particle size distributions, concentrations, and flows are resolved at strategic locations, which make the model suitable as a tool for optimisation and development of next generation systems.
This paper describes the process in the prototype system and how it is modelled, and discusses the simulated results for different system configurations. In addition, field test data of the lamella thickeners cleaning capacity are also presented.

2 Prototype system description

2.1. General description

The process in the prototype system is described in Fig. 2
A plunger pump (P4) pressurises water, which is used for driving the hammer tool and for flushing the borehole. Particle-contaminated water is returned for cleaning before re-use. The cleaning process is based on a lamella thickener with a flocculation system and a hydro-cyclone unit. The equipment was built into a single container (Fig. 3) for ease of transport and handling. The complete system is described in more details by Tuomas (2001). Table 1 summarises some important specifics of the system.

Fig. 2. Schematic flow-chart describing the prototype process.

Fig. 3. Prototype system unit.

Table 1.System specification

2.2. Fluid cleaning system

The prototype cleaning system uses gravity sedimentation for primary separation of particles from drilling water. The lamella thickener is of cross-flow type, leading to a horizontal flow between inclined lamellas (Fig. 4).

Fig. 4. The prototype lamella thickener (T3 in Fig. 2). The unit is of cross-flow type and equipped with screw conveyors for grit discharge.

Particles settle onto the lamella and slide towards the centre of the unit and eventually reach the bottom of the tank. A horizontal conveyor transports the sediment towards the end of the settling unit, where another inclined conveyor removes the waste out of the system. This second conveyor also serves to dewater the waste in order to achieve low water consumption. The settling unit is equipped with a pump for sediment removal if the conveyors are insufficient. Efficiency of sedimentation processes can be significantly improved by adding a flocculent to the incoming slurry flow. These substances gather individual fine and colloidal particles into clumps (flocks) that settle out more easily.
In addition, particle-fluid separation can be achieved with hydro-cyclones. The idea is to use the hydro-cyclones as an alternative to flocculation. The hydro-cyclone unit has a dso cut-point below 5 μm (particles with density 2750 kg/m3 in water). It is designed for a 0.3 m3/min flow and consists of sixty Ø10mm hydro-cyclones.

3. System process model

A numerical model for simulation of particle flows in the prototype system has been developed. Mathematical expressions for significant components are derived, and the whole model is implemented within the Matlab Simulink™ math package. Results of main interest are the time dependent particle size distribution functions Ф(s,t) and corresponding volume flow rate functions, q(t), at different locations in the system. Fig. 5 shows the principle flow scheme and mathematical descriptions of the different blocks are presented in the following sections

Fig. 5. Principle flow scheme of the process model.

3.1. Hammer tool

The hammer tool block in the model adds particles to the system. This is mathematically described as:
Φout(s,t) = Φin(s,t) + Φh(s) (1)
where Фin,(s,t) and Фout(s,t) represent the particle size distributions in the fluid entering and leaving the hammer tool. Фh(s) represent the particles that are generated during drilling. The Ф-functions also represents the volume concentration of particles in the corresponding slurry according to equation:
(2)
where qsolids is the volumetric flow rate of solids and q is the flow rate of slurry. Фh(s) in Eq. (1) is calculated as:
(3)
where v and A represent the penetration rate and borehole cross-area, respectively. Фc(s) is a time independent function which represents the shape of the particle size distribution curve generated by the hammer tool. The curve used in this study (Fig. 6) comes from laboratory analysis of a drill water sample, taken during typical rock drilling on ~ 100 m depth with a 4-inch hammer tool. The shape of Фc depends on various parameters, such as the actual borehole depth, borehole orientation, flow rate, mineral type and drill bit design as well. The shape of the curve is, however, assumed constant in this model. The slurry flow rate, qout, from the hammer block is assumed equal to the incoming flow rate, qin.

Fig. 6. Particle size distribution in a drill water sample, taken during drilling with a 4-inch water driven DTH-hammer tool at approximately 100-m depth. The curve is used to represent function Фc in the described process model.

3.2. Mixing tank and drilling fluid tank

In a tank containing a substance with concentration c, the changed particle concentration by time is described by a differential equation:
(4)
where q is the flow, c is the concentration at n number of intake- and outlet ports in the tank, V is the volume, which may vary with time. After inserting Eq. (2) into Eq. (4), the equation for a tank with n number of intakes is derived as:
(5)
where Фin is the particle size distribution in the fluid entering the tank, qin is the corresponding fluid flow rate to the tank and Ф is the particle size distribution in the tank. The model assumes that both the mixing- and drilling fluid tanks are initially filled-up with clear water. The initial condition to Eq. (5) is, therefore, Ф(s,0) = 0. The volume in the drilling fluid tank will steadily decrease during drilling. The reason is that the separation processes in the lamella thickener and hydro cyclones consume fluid during operation. Opening a water intake at a low fluid level, and closing it when the tank is filled solves this problem. The model is designed to work in a similar way. One of the intake flows, qin in Eq. (5), is changed from zero to a user defined positive value when the tank level V has reached the low limit, and goes back to zero when the upper limit is reached.

3.3. Lamella thickener

The lamella thickener (Fig. 4) is designed for a horizontal flow of slurry between inclined lamellas. Particles settle against the lamella and slide towards the centre of the unit and eventually reach the bottom of the tank. Fig. 7 shows some principle particle trajectories between two lamellas during steady flow conditions. Using symbols in Fig. 7, the critical settling speed for a particle, starting at point (0,y), is calculated by:
(6)
where vcr is the critical settling speed, y is the path start coordinate, U is the slurry flow speed and L is the lamella length. Particles with settling speed lower than vcr, starting at point (0,y), will go to the overflow (accept), while particles with a higher settling speed will end up in the underflow (reject). The actual terminal settling speed for a spherical particle with diameter d is calculated as:
(7)
where d is the particle diameter, g is the acceleration, ρs is the particle density, ρf is the fluid density and Cd is the form drag coefficient. By setting Eq. (6) equal to Eq. (7) and finding the corresponding particle diameter, the efficiency curve for a lamella thickener according to Fig. 7 is obtained. To solve this, an iteration procedure is required, since Cd is a function of the particle Reynolds number, Re, which besides the viscosity depends on the particle diameter and the settling speed.

Fig. 7. Principle outline of two particle paths in a horizontal lamella thickener. Ideally, all particles larger than d2 (belonging to path 2) will go to the underflow. Particles with sizes d

A commonly used function to describe the efficiency of a particle-fluid separation process is the Rosin-Rammler formula (Crowe et al., 1998):
(8)
Here parameter Yc denotes the corrected efficiency and is the probability of a particle with size d to go to the underflow. The parameter d50c is the cut size of the corrected grade efficiency curve (or corrected partition curve), and m is a factor that affects the sharpness of the curve. When the underflow is taken into account, the efficiency Y is calculated by:
Y= (1-R)Yc+R (9)
where R is the fraction of incoming fluid that goes to the underflow. Several conditions must be fulfilled for Eq. (9) to be useful in estimating the separation efficiency in a lamella thickener. The following simplifica¬tions and assumptions have been made to motivate the use of the equation:
1. Even though the slurry flow is discontinuous due to drill pipe installations, the flow is periodic with constant run/stop times. The values of parameters d50c and m can thereby be chosen so that the curve Y represents the mean separation efficiency during a complete period.
2. The flow rate deviation of the slurry is small and does not affect the separation efficiency.
3. The amount of solids in the slurry is approximately constant and deviations do not affect the separation efficiency.

Table 2. Parameters and data used in numerical simulations

Table 3. Description of simulation run A, B and C

Simulation A	Particle separation by the lamella thickener, without the hydro-cyclones.
Simulation B	Particle separation by the lamella thickener and the hydro-cyclones. Underflow from the hydro-cyclones are disposed
Simulation C	Particle separation by the lamella thickener and the hydro-cyclones. Underflow from the hydro-cyclones are fed into the mixing tank.

The actual values of the parameters dSOc and m are chosen with respect to field data (Fig. 10) and calculations according to Eqs. (6)-(9). R is adjusted individually in each of the following simulations but is close to 0.1. The reason is that the experience based amount of ~50% w/w solids in the underflow should be fulfilled. The value for the overflow and underflow are calculated by equations Eqs. (10) and (11):
q0 = q(1-R) (10)
qu = qR (11)
where index o and u denote overflow and underflow, respectively. Particle size distribution curves are calculated by:
(12)
(13)
where Фu(s,t), Ф0(s,t) and Ф(s,t) represent the particle size distributions in the underflow, overflow and the feed. The terms (1/R) and (1/1-R) are required for the functions to correctly represent the new concentration levels together with the new flow rates.
Another detail to consider is that the volume of the lamella thickener causes a delay of Δt=v/q s before the incoming slurry flow particles are reported in the overflow or underflow. This is the case when the flow is laminar and no mixing occurs. The described model uses a memory buffer to stall the signal Δt s, according to Eq. (14):
Фout(s,t) = Фin(s,t – Δt) (14)
Фout(s,t) = 0 для t≤Δt (15)
Eq. (15) implies that the lamella thickener is initially assumed filled with clear water.

3.4. Hydro-cyclones

Several equations for calculation of the separation efficiency for hydro-cyclones have been derived during the past decades (Heiskanen, 1993). The parameters involved are the geometry, operating conditions and fluid characteristics. One model that is often used is the empirical model given by Plitt (Wills, 1997):
(16)
In this equation Dc = hydro-cyclone diameter [cm], D0 = overflow diameter [cm], Di = inlet diameter [cm], φ = volumetric fraction of solids in the feed, Du = underflow diameter [cm], h = cyclone height [cm], Q = feed volume flow rate [m3/h], ρs = solids density [g/ cm3] and ρf= fluid density [g/cm3]. The sharpness parameter m is given as:
(17)
By inserting Eq. (16) and Eq. (17) into the Rosin-Rammler formula Eq. (8) and compensating for the underflow Eq. (9), the efficiency curve for the hydro-cyclone unit is obtained.

Table 4.Results from numerical simulations

4. Resulting calculations

The described model has been implemented within the Matlab Simulink math-package and three different simulation runs representing possible system configurations are presented here. Input parameters and conditions are described in Tables 2 and 3. Results are presented in Table 4 and Figs. 8 and 9.

Fig. 8. Results from numerical simulations that describe the volume concentrations of solids in the flow to the hammer tool. Curves A, B and C result, respectively, from simulation runs A, B and C.

Fig. 9. Results from numerical simulations that describes the particle size distribution curves at t=24000 s (Fig. 8) in the flow to the hammer tool. The area under the curves represents the volume concentration of solids in the flow.

4.1. Results

Important results from the simulations are the total particle volume sent to the hammer tool and the external water consumption. The reasons are that the hammer life is intimately related to the presence of abrasive particles in the flow and that water consumption (and thereby waste flow) must be low for efficient use of the system. Results presented in Table 4 indicate that the mean concentration of solids in the feed to the hammer tool is approximately 0.44% w/w when the lamella thickener is used for particle-fluid separation (simulation A). Particle flow is reduced by ~ 80% when hydro-cyclones are used as a complement (simulation B) and the underflow is disposed. When underflow from the hydro-cyclone unit is re-used (simulation C), the reduction is about 50%. The simulations are valid for the case of no flocculent in the flow. Fig. 8 describes the solids volume concentration curves in the flow to the hammer tool, during drilling of a 200-m deep borehole. The particle size distribution curves (for t=24 000 s in Fig. 8) are presented in Fig. 9.

5. Conclusion

Drilling with water driven DTH-hammers is a recently developed method for competitive production of boreholes. The technology requires large flow rates of preferably high quality fluid to drive the hammer tool and flush the borehole. One method to reduce the consumption is to process and recycle the used drilling fluid. Studies have been performed to find cost-effective suitable cleaning methods and a mobile prototype unit has been developed. This unit includes components for both pressurising and cleaning drilling fluid to enable recycling and thereby efficient drilling.
A process simulation model was also developed within this project. Simulations determine particle size distributions, concentrations and flows at strategic locations, whereby the system configuration can be optimised. Results indicate that particle flows to the hammer tool is reduced by ~ 80% when hydro-cyclones are used as a complement to the lamella thickener, and the underflow is disposed. When underflow from the hydro-cyclone unit is re-used, the reduction is about 50%. The simulations are valid for the case of no flocculent in the flow.
Practical prototype experiences and results from numerical simulations will be used in designing next generation systems, leading to even more cost-effective production with further increases in the competitiveness of the drilling method.

Acknowledgments

This work was supported by Technology Link Foundation, The Research Council of Norrbotten and Wassara AB. They are greatly acknowledged.

Appendix A: Field data

The mining company LKAB in Malmberget, Sweden, has during year 2001 produced several boreholes for safety investigations. The distances from the ground down to the mine were measured and rock surveillance systems were installed to monitor movements. The latter is a sign of instability and, therefore, a hazard for the residents and the surrounding environment. The holes were drilled using water driven DTH-hammer tools together with the above described prototype system. Recycling was used during approximately 200 m of drilling, and the capacity of the lamella thickener was studied. Conditions for this drill work are presented in Tables 5 and 6. Laboratory results are shown in Table 7 and Fig. 10.

Table 5. Operating conditions during field experiments

Table 6. Time and positions for samples

Fluid sample no.	Position	Time (min)
1	Lamella thickener overflow	30
2	Lamella thickener overflow	60
3	Lamella thickener overflow	90
4	After hammer tool, before addition of flocculent	0

Table 7. Laboratory results from flow samples

Flow sample number	Mass fraction solids (%)
1-overflow	0.05
2-overflow	0.10
3-overflow	0.05
4-feed	8.4

Fig. 10. Laboratory results of particle distributions in fluid samples. Samples 1—3 are taken from the lamella thickener overflow at time 30, 60 and 90 min, respectively. Curve 4 describes the feed at t=0 and before addition of flocculent. The flow rate was ~0.21 m3/min.

References

Crowe, C., Sommerfeld, M., Tsuji, Y., 1998. Multiphase Flows With Droplets and Particles. CRC Press.
Heiskanen, K., 1993. Particle Classification. Chapman and Hall.
Tuomas G., Nordell B., 2000. Down-Hole Water Driven Hammer Drilling For BTES Applications. In: Proceedings Terrastock 2000, 8th International Conference on Thermal Energy Storage. Stuttgart, Germany, pp. 503-508.
Tuomas G., 2001. System for Water-Driven Downhole Hammer Drilling. In: Proceedings Offshore Technology Conference OTC 2001, April 30 to May 3, Houston, Texas, USA, pp. 399-407.
Wills, B.A, 1997. Mineral Processing Technology. Butterworth-Heinemann.
Oderyd, L., 2001. Personal communication.