Andy Smith
MIne-action specialist

Comparative trials of manual mine clearance techniques, Mozambique

December 2004

All photographs and diagrams copyright Andy Smith 2004 unless otherwise acknowledged.



This report was originally compiled by Andy Smith for submission to GICHD as part of their Manual Demining Study 2004. It was compiled without payment and outside any contract timeframe, so remains the property of the author. This version has not been edited or approved by GICHD, and so GICHD bears no responsibility for errors or omissions of fact or opinion that may occur in this document. GICHD made changes and used parts of this document in their Manual Mine Clearance Study report.

Accelerated Demining
Programme ( ADP )

Without the committed assistance of ADP management and staff, these trials could not have been conducted.

The trial was devised and implemented by Andy smith (AVS), with valuable advice and input from all persons/organisations participating in the planning, monitoring and evaluation, especially the representatives from BAM and QinetiQ:

Jacky D’Almeda, Florencio Chongo: Manuel R. Django, Paulino Tule Gove: ADP Mozambique
Mate Gaal, Christina Mueller, BAM
Neville Goulton, Dai Lewis: QinetiQ
Jan Eric Stoa, Luke Atkinson: NPA Sri Lanka
Sara Sekkenes, Hans Rune Kampenhøy: NPA Mozambique

Tim Lardner, Havard Bach: GICHD
Dieter Guelle

The following ADP field supervisors and deminers worked selflessly to support the trials:
Fernando Laice, Dionisio Chaka, Julio Ernesto Wache, Carlos Tembe, Januario Justino Mindo, Helder Martins Adido, Manuel Abilio Balate.


Terms, definitions and abbreviations

Familiarity with acronyms commonly used in Humanitarian Mine Action is presumed.

area-excavation: in this report, the term “area-excavation” is used to describe the process of removing the entire ground surface to a predetermined depth, and locating any concealed mines or ERW in the process.

detector-signal investigation: in this report, the term “detector-signal investigation” is used to describe the process of locating metal with a metal-detector, then unearthing and recovering that metal from a discrete location.

REDS – Rake Excavation and Detection System, as devised by NPA with the HDU in Sri Lanka.

GRH – Ground Reference Height: a measure of electromagnetic disturbance from the ground.


1 Foreword

2 Introduction

3 Comparative trials in Mozambique

3.1 Aims

3.1.1 Limitations to aims

3.2 Trial conditions

3.2.1 Total trial area

3.3 Trial duration

3.4 Targets

3.5 Recording data

3.6 Trial Monitor’s duties

4 Manual demining systems under assessment

4.1 Standard method (Trial 1)

4.2 Standard plus magnet-clip on tool (Trial 2)

4.3 Standard plus MBR (Magnet Brush-rake) – Trial 3 .

4.4 Detector in low-fragment area (Trial 4)

4.4.1 The “reliability” part of Trial 4

4.5 The REDS rake system (Trial 5)

4.6 Standard ADP spade excavation (Trial 6)

4.7 Standard NPA (Mozambique) excavation (Trial 7)

4.8 Standard Mattock excavation (Trial 8)

4.9 Prodding (Trial 9)

5 Results

5.1 Results: Standard metal-detector method (Trial 1)

5.2 Results: Standard plus magnet-clip on tool (Trial 2)

5.3 Results: Standard plus MBR (Magnet Brush-rake) – Trial 3

5.4 Results: Detector in low fragment area (Trial 4)

5.4.1 Results of the “reliability” part of Trial 4

5.5 Results: The REDS rake system (Trial 5)

5.6 Results: Standard ADP spade excavation (Trial 6)

5.7 Results: Standard NPA (Mozambique) excavation (Trial 7)

5.8 Results: Standard Mattock excavation (Trial 8)

5.9 Results: Prodding (Trial 9)

6 Conclusions

6.1 AVS Conclusions from the results

6.2 AVS Recommendations

6.2.1 AVS Lessons learned for future trials

6.3 Conclusions fom observer organisations

6.3.1 BAM conclusions from the results

6.3.2 BAM Lessons learned for future trials

6.3.3 QinetiQ conclusions from the results

6.3.4 QinetiQ recommendations

Annex A: References

Annex B: Trial records AVS, November 2004

Annex C: Debriefing of Deminers and Section-Commanders

Annex D: QinetiQ trial report [Not reproduced]

Annex E: BAM trial report [Not reproduced]


A series of comparative trials of manual demining techniques were undertaken at Moamba in Southern Mozambique , with essential assistance from (UN) ADP . These limited trials pitted varied manual demining techniques against each other in a common environment in order to assess their relative efficiency in terms of speed and safety. Various demining groups were invited to participate, and NPA in Mozambique , NPA in Sri Lanka and ADP Mozambique provided essential manpower, training and monitoring assistance.

To maximise the quantitative output that would be of statistical value, the German institute BAM advised during the planning phase. To maximise the objectivity of the analysis of the results, the British research group QinetiQ, BAM and ADP QA staff assisted with the monitoring of the trials. All, including the deminers themselves, produced analyses of what could be derived from the trials. Some of the results are qualitative, and some quantitative. When all results coincide, the level of confidence in the accuracy of that result is believed to be unusually high.

Eight manual demining techniques were compared in terms of speed, safety to the deminer, safety to the end-user, comfort and confidence. These methods were:

      1. standard metal-detector use with signal-investigation tools
      2. standard metal-detector use with magnet attached to signal-investigation tools
      3. standard metal-detector use with signal-investigation tools and a “magnet Brush-rake”
      4. area-excavation using an enxada (mattock) and conventional investigation tools
      5. area-excavation using a conventional garden spade and conventional investigation tools
      6. area-excavation using an NPA excavator and conventional investigation tools
      7. area-excavation using a rake-based system (REDS)
      8. prodding, using standard low-friction prodder and conventional investigation tools

The standard use of metal-detector with magnetic signal-investigation tools was repeated with and without excessive fragmentation in the area, and a separate trial to determine the accuracy of deminer pinpointing was also conducted.

The trial area was set up with identical test lanes from which all undergrowth had been removed. The lanes contained concealed mine surrogates that accurately reflected the size and detector signature (to the Minelab F1A4 that was used) of Type 72 AP blast and GYATA-64 AP blast mines. Randomly placed in a manner that ensured that the deminers did not know how many were in an area, the surrogate mines included a witness plate on the top surface to record damage that occurred during their recovery.

The trials compared entire systems, not just the tool variations. The systems included area-marking and internal QA from the supervisors of the deminers. For all systems, the required clearance depth was the Mozambique National standard of 13cm.

All systems except prodding were effective at locating mines, although some deeply buried mines were missed. In most cases, those missed were Type 72 surrogates buried at 12cm to the top of the mine (about 13cm to the top of the metal insert).

In a heavily fragmented area, the most efficient method of clearance was using a metal-detector and a magnet Brush-rake. The use of signal-investigation tools that included a magnetic attachment was the next fastest. The use of a magnet Brush-rake in areas with cut vegetation or leaf litter might have given an even greater speed advantage.

When a metal-detector was NOT used, the most efficient method of clearance was that involving the use of a conventional garden spade and conventional investigation tools.

The method most likely to involve an accident to the deminer carrying it out was prodding. Prodding at 30 ° to the ground achieved an average clearance depth of less that 4cm, and all the mine-surrogates that were located during the trial had been damaged by prodding onto their pressure-plates.

After prodding, the method most likely to involve a deminer accident was area-excavation using a mattock (enxada) (this finding coincides with that derived from the available accident records in the DDAS - UNMAS/GICHD Database of Demining Accidents, DDAS, 2005).

The methods most likely to leave mines behind were area-excavation in which the required clearance depth was not maintained, and excessive speed meaning that deep signals were missed during metal-detector based clearance.

A post-trial evaluation of the methods and approaches employed during the trials determined that the variations between different demining teams meant that the different tools and methods were not the only variables affecting the results. An expanded repetition of the trials that allowed each separate demining team (deminers and supervisor) to use each method in turn would provide a more direct comparison, and so yield more compelling results.


1. Foreword

This report was originally compiled by Andy Smith for submission to GICHD as part of their Manual Demining Study 2004. It was compiled without payment and outside any contract timeframe, so remains the property of the author. This version has not been edited or approved by GICHD, and so GICHD bears no responsibility for errors or omissions of fact or opinion that may occur in this document. GICHD used parts of this document in their Manual Mine Clearance Study report 2005.

2. Introduction

Following AVS field studies that identified and confirmed common techniques, tool and processes employed in manual demining, formal field trials of a variety of manual demining techniques were devised. Partners with specialist skills in research and in statistical analysis were invited to take part in the trials, and the trials were conducted with assistance from three field demining groups in Mozambique during October and November 2004. The leading field partner was ADP Mozambique, which provided monitoring and evaluation staff as well as deminers, equipment and a wide variety of other resources.

The purpose of the evaluative trials was to compare the relative efficiency of varied manual demining techniques. The trials were conducted at a training base belonging to ADP and situated in rural Mozambique. The trial area was prepared in a manner designed to limit variables and ensure that each method under trial was assessed in a context and under circumstances that were as similar as possible, and which closely reflected demining reality in that region.

Quantitative and qualitative evaluation of the results was conducted by QinetiQ UK, BAM Germany, the ADP deminers and the author of this report. Where the varied conclusions derived from the trials coincide, it is felt that the variety of skills and experience held by those assessing the results adds compelling weight to both the quantitative and qualitative findings.

The evaluation reports provided by QinetiQ and BAM are given in Annex D and E of this report. [Respecting their copyright, these separate reports are not reproduced here.] The results of interviews with the deminers are reproduced in Annex C.


3. Comparative trials in Mozambique

Trials of manual demining methods took place in Moamba, Mozambique during October and November 2004.

Of those trial categories listed under IMAS 03.40, these trials are best described as “concept and technology demonstrator trials”, although unanticipated outcomes gave them something in common with “demonstration” and “acceptance” trials.

The most common system of manual-demining - using a metal detector and signal-investigation tools - was compared with selected other manual demining techniques involving a variety of tools.

The organiser and supervisor of the trial was Andy Smith. The NMAA in Mozambique ( IND) was invited to attend. The main partner and trial facilitator was ADP in Mozambique. While the responsibility for the design and conduct of the trials rests with Andy Smith, he acknowledges essential support and input in trial design from BAM, and in the conduct of the trials from ADP, QinetiQ and BAM.

3.1 Aims

The principal purpose of the trials was to provide quantitative and qualitative data allowing a comparative assessment of the “efficiency of selected manual demining systems. An informed qualitative overview would then be applied to the results.

“Efficiency” was to be measured in terms of speed, the location of targets within a predefined depth, the safety of the deminer while conducting the varied drills, the safety of the end-user of the land, deminer comfort and deminer confidence.

The trial of complete manual demining “systems”, including familiar PPE, marking, tools and QA checks, was intended to limit the number of “unfamiliar” variables that the deminers had to deal with, and so make it more likely that the variation in results was a direct consequence of the particular variation under assessment.

3.1.1 Limitations to aims

It was recognised that the trials could only be conducted in one place at one time, and with a small set of experienced deminers, so the “findings” could not always be broadly applicable.

There were strict time and cost constraints on the trials, with a non-negotiable finish date. The risk of inclement weather was high, so whenever possible, trials were conducted in parallel to make maximum use of fine-weather days.

The number of experienced monitors was limited, and very little time was available for instruction and practice. The need to use inexperienced monitors meant that the quality and quantity of information recorded for each trials was expected to vary. The core data requirements were deliberately limited in order to ensure that the essential data was successfully gathered.

The combined constraints meant that it was not realistic to aim to “prove” anything definitively. The objective was to achieve consensus over conclusions that, in context, could be compelling.

3.2 Trial conditions

The trials were conducted in selected parts of the ADP test area in Moamba, near Maputo in Southern Mozambique. All trial areas were prepared in an "identical" manner, although the position of the concealed surrogate-mines varied (the positions were randomised using computer software).

All trials were conducted in areas where the vegetation had been cut and lanes marked. Surrogate-mines and typical minefield metal fragmentation had been concealed in mapped positions inside the lanes, with the position of the targets determined by the statistical specialists at BAM with a view to leaving the deminers uncertain about the number, depth and position of surrogates that were placed in any one lane.

The Ground Reference Height (GRH) throughout the trial area was measured by BAM, and reported to have been less than 10cm throughout the area, with no significant variations.

Surrogate-mines were prepared with a metal content that made the Minelab F1A4 metal-detector signal with the same strength at the same depth as when a real mine was used. The surrogates were concealed at two depths, 1cm and 12cm to the top of the target.

Experienced deminers worked to approximately their normal work/rest routines and under the supervision of a "familiar" Section Commander. Their actions were recorded by a Trial Monitor who was constantly present. The Trial Monitors observed and recorded events without interfering with the work in progress.

During the trials, the clearance depth required was Mozambique's minimum standard of 13cm to the top of the mine/device.

Each test was conducted over an area comprising four 5 metre long and one metre wide lanes. The lanes were separated. This kept the marking requirement identical and meant that any targets near the side of the lane had to be found during that lane's clearance, not the next. It also allowed full observation without physical interference.

Two test areas (each of two lanes) were used for each trial, which was conducted by two "one-man-drill" deminers and a single field supervisor (Section Commander) controlling both deminers and carrying out routine internal QA checks on their areas.

3.2.1 Total trial area

It was decided to use a “herringbone” design for the trial area so that deminers engaged in the same trial could work close to each other while having their backs to each other and so being unaffected by the other’s progress.

Two strips 120 metres long and seven metres wide were prepared. The area between these strips was 20 metres wide. Inside each of the strips, 18 five metre long lanes were measured and marked out, resulting in a total of 36 separate five metre long lanes.

In each strip, the lanes were grouped in pairs with a metre separating the pair, and five metres between it and the next pair [in one case this was varied to avoid an animal path].

One deminer used a pair of 5 metre lanes in each Trial. The second deminer in that Trial used the pair of lanes opposite him in the “herringbone”. One Section Commander oversaw the work of both deminers, much as he would have in a live area. One Trial Monitor could then record the necessary data from both deminers if required, although there was a Monitor for each deminer during almost all of the trials.

3.3 Trial duration

All trials (except 9) used two deminers and a supervisor for up to three days or ten metres cleared, whichever was sooner. This meant that there were time constraints as well as an ultimate area constraint of 20m per trial.

The total time period for setting up, training, conducting and closing down the trials was 37 calendar days. The trials themselves were conducted over 15 days (three working weeks).


4. Targets

Surrogate mines of the right size and colour were designed and made. The numbers were 75 GYATA-64 and 75 Type 72 AP blast mines. These mine-types were selected because they represented a familiar threat to the deminers taking part in the trials. Their preparation and placement only took place in the week before the trials started. Vertical holes were made for placement and water was used during the infill to help the ground settle and disturbed grasses to root. The extreme dryness of the ground led to the rapid spread of the water and helped to effectively conceal the position of the surrogate mines so that the deminers proved unable to see or predict their position, even while digging towards them from the side.

The picture above shows how the depths were measured and all surrogate mines placed “level” horizontally.

Each surrogate mine included a “witness-plate” achieved by applying a thick latex (rubber) with a matt varnish overlay to the top of the surrogate. If struck from above, the varnish cracked as the latex flexed. When struck with enough force, the latex was cut and when the tool was withdrawn, the latex “self-sealed” to preserve the evidence of the impact. This worked very well.

The picture above shows GYATA-64 surrogates being prepared. The white latex dried clear.

The selection of a metal content for the mines followed a pattern devised by Dieter Guelle in previous detector-performance trials. The Minelab F1A4 detector was used with real mines in the same test area, and a metal insert that gave a similar signal at the required depths was selected for insertion inside the surrogate mines.

The surrogates were not accurate replicas, but they looked convincing, and were all identical, so were constant throughout the trials.

In addition to target mines, approximately identical collections of scrap fragments (from local minefields) were placed for those trials where metal-detector use was being compared with area-excavation . Fragments were positioned flush with surface, and at 1 cm and 2cm depths. The proportion of ferrous to non-ferrous fragments was 30:1, so exceeding the proportion found in the local mined areas (50:1).

The picture above shows three approximately identical collections of mined-area fragmentation, including bullets, barbed wire and typical mined-area fragmentation for that region. (Most fragments were from OZM series bounding fragmentation mines.)