Last month we looked at the reasons why, as geoscientists, we need low frequencies. We also reviewed the sensors used to receive the reflected seismic signals and the recording instruments. This month’s article will address some of the issues of the seismic source. In other words, how do we get the required low-frequency energy into the Earth? The two most frequently used sources for land acquisition are vibrators and explosives. We will assess the operational and cost differences between them and the advantages and disadvantages of each. In addition, a quick reminder: when considering the bandwidth of seismic data, we need to think in terms of octaves rather than in terms of hertz.
Discussion: Explosives
Most geoscientists will remember early lessons about seismic exploration in which they were taught that dynamite is used to create an impulse or sharp spike in the ground, and that this energy would travel through the Earth’s layers to be reflected back to the surface. The reflected energy would then be recorded at the surface and carefully processed to produce an image of the structure below.
Unfortunately, this is not reality.
First, the seismic industry has not used pure dynamite for a considerable number of years. Modern seismic explosives might use some nitroglycerin, but in combination with other chemical additives to improve the safety (stability, toxic gas emissions, etc.) and performance (detonation velocity, gas volumes, etc.). Other explosives used might not contain any nitroglycerin whatsoever.
The second issue is that explosives do not produce a perfect spike. It has been well known for many decades that, when an explosive charge is detonated, it creates a gas cavity surrounding the explosion. The rock near the charge is fractured and the gas expands into these fractures. At a small distance (typically less than 2 meters) from the actual explosion, the fracturing of the rock stops and the motion becomes elastic. The gas within the cavity and the elastic nature of the rock cause reverberations or oscillations, not a spike. The oscillations have a resonant or dominant frequency that depends primarily upon the charge size and the elastic properties of the surrounding rock. The detonation velocity and gas volume have a lesser effect.
Wim Peet, in 1960, derived simple equations that show the relationship between the resonant frequency and charge size. Figure 1 shows an example of Peet’s equations with well-tamped charges in uniform and competent clay. Here we can see that a 1-kilogram charge in this clay has a dominant frequency of about 30 hertz. For very low dominant frequencies, the required size of the charge increases very rapidly. Simply put, to obtain a low frequency dominant frequency we need a large gas cavity and therefore a larger charge.
Although the explosion has a dominant frequency, the recorded data will still contain signal with both higher and lower frequencies, which can hopefully be recovered with good data processing. Figure 2 shows a comparison of amplitude spectra that were averaged over three shots each, of 1-kilogram and 40-gram charges, spaced along a 2-D seismic line. The well-tamped charges at each shotpoint were planted in 24-meter-deep holes, 2 meters apart, and the spectra were measured over 1-second windows for 750 meters on either side of the source locations. The data were recorded using MEMS (micro-electro mechanical system) accelerometers and integrated to provide data in the velocity domain. As can be seen in the main spectrum plot, the 1-kilogram shots show their highest amplitude at about 22 hertz, while the small 40-gram charges have a much higher frequency resonance at about 85-90 hertz. At frequencies greater than 80 hertz, the smaller charge is generally about 10 decibels higher amplitude than the 1-kilogram charge. It is interesting to note that in this instance, the dominant frequency of the 1-kilogram charge is lower than that shown in figure 1 (about 22 versus approximately 30 hertz). The charges in these recordings were in soft, poorly compacted soils in which the elastic properties were less stiff and the gas cavity could expand further, resulting in lower frequency resonance.
If we look carefully at the inset window, zoomed over the lower frequencies, there are several important observations that can be made. In the range from 10 hertz to about 22 hertz, the 1-kilogram charge shows amplitudes that are 12 decibels larger (as in, four times greater amplitude) than the smaller one. At the lower frequencies there is a flattening of the spectrum at about 7 or 8 hertz on the 40-gram charge and 5 to 6 hertz on the larger charge. This is interpreted as being the noise floor. For recovery of lower frequencies than these in data processing, very high trace density recording would be needed. When using explosives, if increasing the trace density requires a corresponding increase in shot density, this would add significantly to the acquisition cost. At frequencies below 1.5 hertz we can see that both spectra are beginning to increase in amplitude. This is an indication of the thermal electronic or Johnson noise that was discussed in last month’s article.
From an environmental standpoint, large explosive charges necessarily require deeper shotholes to prevent blowout. From an economic point of view, the cost of drilling deep holes can be extremely high. In areas where explosives are operationally the best option, this has led to designing surveys with closely spaced smaller charges for the shallow and intermediate depths, and widely spaced larger charges to provide improved velocity modeling and deep data. In areas where seismic acquisition is feasible using vibrators, these have become the source of choice for obtaining both low and high frequencies.
Discussion: Vibrators
Although there are many seismic crews around the world still using explosives, if we consider the actual number of source locations, then vibrators are by far the most common seismic source. All seismic vibrators work on the principle of Newton’s second law of motion: Force equals mass times acceleration. As frequencies approach 0 hertz, inertial mass acceleration approaches zero, and the force rapidly decreases. For more force, we need either a lot more mass, or a greater range of motion, so that acceleration can be greater. Conventional wisdom has been that we need a larger, heavier machine with a larger engine and hydraulic oil pump to get more force at low frequencies. Over many years we have seen gradual increases in the size and weight of seismic vibrators to the point that the largest of these are no longer street-legal in many regions. Operational constraints often limit their use to deserts, open plains and the Arctic.
Counter rotating eccentric mass vibrators can produce very low frequencies with extremely low distortion but have not been used in commercial operations since about 1960. Their limitations include severely limited high-frequency bandwidth, large size and large weight. Electromagnetic vibrators are also capable of producing very low frequencies with extremely low distortion but have not yet evolved into high power units. I expect to see more of these in the near future.
Since existing servo-hydraulic vibrators are limited in their ability to produce adequate force at very low frequencies, in recent years we have seen the introduction of “low-dwell” sweeps. In these, additional time is spent sweeping through the low frequencies at the start of the sweep to compensate for the lower force output of the vibrators. Figure 3 shows an example of three 30-second sweeps with start frequencies of 3, 2 and 1 hertz. The time spent at each frequency is adjusted to maintain a flat amplitude spectrum input into the Earth but, of course, the overall force at any single frequency is reduced. Once the maximum force output of the vibrator has been attained, the sweeps generally change to a linear function of frequency versus time. On the left is a 3 to 80-hertz sweep showing the reduced amplitudes and extended time at the lower frequencies. In the middle, the sweep is 2 to 80 hertz and the time spent in the extra 1 hertz is about 3.5 seconds. On the right is a 1 to 80-hertz sweep. The time spent to acquire the frequencies from 1 to 2 hertz is approximately 16.5 seconds, which is more than half the length of the sweep. In order to maintain the same signal-to-noise ratio at the higher frequencies as with sweeps not containing the additional low frequencies, additional sweep effort would be necessary, using either more or longer sweeps. Remember the comment in last month’s article: “There is frequently a substantial increase in data acquisition cost that results from acquiring this additional octave.”
Harmonic distortion has been a serious problem with servo-hydraulic vibrators, particularly at the lower frequencies. Significant improvements have been made in the mechanical construction of the vibrators, the hydraulics, and the electronic control systems. These have allowed coherent low-noise, low-frequency signal to be generated with vibrators. The source used for figure 5 in last month’s article was acquired using a low-dwell sweep with a large, heavy vibrator.
A relatively simple method of improving vibrators’ output force at low frequencies, which has been overlooked until recently, was announced at the most recent Society of Exploration Geophysicists annual meeting. Instead of adding weight to the inertial (reaction) mass or increasing its range of motion, we can “borrow” a fraction of the carrier vehicle’s mass, when needed, to generate more force at low frequencies. A concept verification prototype supports the idea, and development is continuing. During the low-frequency part of the sweep, mass is temporarily transferred from the vehicle to the inertial mass through hydraulic dampers activated with electrical signals. Figure 4 illustrates the potential boost in low-frequency force attained by modifying an existing vibrator. Existing vibrators can be retrofitted for a small fraction of a new vibrator’s cost. Patents for this latest technology are pending for Servo Force, LLC.
Conclusions
The generation of adequate strength, very low-frequency signal by seismic sources is difficult and potentially expensive. Use of explosives requires the use of large charges with a corresponding increase in the cost of drilling and of the environmental footprint. With current vibrator technology, the use of low-dwell sweeps has successfully enabled low-frequency data acquisition in a number of areas. In many of these areas, the increase in data acquisition time caused by the additional sweep time necessary has been minimized by use of large fleets of individual vibrators sweeping simultaneously. Unfortunately, in congested urban areas, jungles or rough topography, these applications are much more difficult to accomplish.
With implementation of the vibrator technologies described above and using conventional 10-hertz geophones, we should be able to recover frequencies down to 2 or 3 hertz. Using geophones with lower resonant frequencies (5 hertz, for example) we should be able to get close to 1 hertz. With typical explosive charge sizes and 10-hertz geophones, perhaps 3 to 4 hertz. With low noise MEMS and careful survey design, equipment and parameter selection, we should be able to recover frequencies down to between 1 and 2 hertz.
Acknowledgements
The author would like to thank Dennis Reust of Servo Force, LLC for the information related to the vibrator technology released during the 2020 SEG annual meeting and the graph shown in figure 4.