High precision GPS guidance systems on dozers can locate the position of the tracks or the blade within a few centimetres to form a continuously updated Digital Terrain Map (DTM) of the job.
As the job progresses it is possible to use the data to calculate volumes of material moved.
In figure 1 the report shows productivity for a single dozer over the course of an 11-hour shift.
From 7am to 8am the system was on for 56% of the time. Since the system comes on when the dozer starts, this means the operator started the machine at 7.25am.
The second column shows the machine was idle for 10% of the remaining time. In this report, idle was defined as any 30-second period when the dozer moved no more than 0.5m. If the dozer is not moving it is not doing anything productive.
The third column shows the machine moved 495m3 of cut.
The fourth column shows 387m3 of fill. In theory the cut should match the fill after allowing for swell. There are many reasons why cut will not match fill in an individual hour. In this case the dozer was pushing over an edge. The system can only measure what the dozer runs over, however, the cut number is accurate.
The last three columns show the cut was done in nine pushes, with an average length of 123m up a slope of 1.4%.
A common method of calculation, used in the example above, is to divide the job area into a grid. By keeping track of the elevation of each grid square as the dozer passes over it, volumes moved can be tracked. The size of each grid square is important.
A compromise is necessary between small grid dimensions that improve accuracy and large dimensions that reduce the amount of data handled. I have found a grid size equivalent to one quarter of the dozer blade gives good results.
The data stream from the dozer consists of its 3-D position at an instant in time. Again a compromise is necessary between accuracy and excessive data. A reporting interval of four seconds has proven satisfactory. In this time the dozer will cover about 10m. The elevation each grid square is interpolated from successive positions.
Using grid squares, the volume calculation concept is simple. Whenever a grid square shows a reduction in elevation, then the area of the grid square times the reduction in elevation gives the volume of cut. Whenever a grid square shows increased elevation, a volume of fill can be calculated.
This concept is simple, but the situation is more complicated when grids previously cut are filled and vice versa.
In order to give a true report of productivity, you have to properly account for rehandle.
Consider the extreme case where the dozer excavates a 100m3 hole and stacks it on the surface, and that the material swells by 10%. At this point the report will show 100m3 of cut prime and 110m3 of fill prime.
Suppose now the dozer pushes the material back into the same hole. I suggest the report for the entire period will show: Prime cut = 0m3 - Prime fill = 10m3 - Rehandle cut = 100m3 - Rehandle fill = 100m3
In essence, a lot of work has been done but no production.
If the prime movement took place in the first hour and the rehandle in the second hour, then the first hour report will have to be adjusted after the second hour.
This later adjustment of earlier reports happens in practice, is counter intuitive and is also absolutely correct.
As well as the legitimate rehandle discussed above, early field-testing returned very large volumes of rehandle. Investigation of these numbers shows several possible reasons:
When pushing parcels of material for considerable distances, the exact track followed can vary a little. Continual small movements up or down can add significantly to rehandle. Setting the system to only react to movements of 150mm or more corrects this issue.
When the dozer runs through a low point, as in dozer stripping for example, capturing a position at set time intervals may give wrong results. For example, on one pass the dozer may capture the position of the lowest point, whereas on the next pass it captures points on either side. The machine is apparently cutting and refilling a triangle. Setting the machine to report a new point whenever the pitch of the dozer changes significantly overcomes this problem.
The GPS very occasionally gives results hundreds or even thousands of metres in error. The GPS hardware is reporting positions five to 10 times a second. Normally a single GPS wrong position goes unnoticed. On a machine guidance system it will cause a temporary flicker of the screen.
However, on captured data, such a glitch may result in a huge amount of phantom rehandle. The software is set to filter out such glitches.
In order to consistently report positions it is necessary to know the orientation of the dozer as well as its location.
A single GPS receiver can only give a position. It is also necessary to know the direction of travel of the machine. Orientation can be determined by using a second GPS antenna and receiver, which is expensive; using a compass, which is not very accurate; or using successive positions to determine direction, which involves software complications but works well and does not add further cost.
When multiple dozer fleets are operated on one job, it is essential to separate the productivity of different machines. It is possible to do this in the software provided the position data is time stamped and all clocks in the system are synchronised.
Ideally the data for the time stamp should come from the GPS.
Besides giving position data, the data from GPS satellites can be used to provide a highly accurate time signal.
Transmission of position data for productivity monitoring requires a lot of data capacity. It is possible to run up to seven dozers on UHF data radios running at 38KB, but the higher bandwidth of wireless Internet systems is preferable.
The usual frequency of wireless systems (2.4GHz) means contact must be line of sight. In order to keep the repeaters to a manageable number, a meshed system is necessary where each dozer acts as a repeater.
The system is more useful if downtime can be properly reported. On Automated Positioning Systems ’ products the system asks the operator the reason for the delay after a machine has been immobile for a user specified period – usually 30 seconds. The operator is offered a menu of possible delays, such as crib or refuelling.
APS has installed monitoring systems on a number of sites involving dozer stripping and dozer mining of coal.
At Yallourn Energy the system has been in use for routine reporting for almost three years.
The same concepts can be used on other mining and earthmoving equipment. On some items such as scrapers, it is relatively easy to use the same methods.
*Chris Seymour is MD of Automated Positioning Systems.