1 Current situation
Gold has been recovered from fluviatile sediments for thousands of years. The saga of the Golden Fleece, for example, has its background in the use of sheep fleece to trap gold spangles in the streams of the Western Caucasus. Together with other high-density minerals (heavy minerals), gold is often present in the form of fine particles (hereinafter referred to as gold spangles) as a minor constituent in gravel and sand, and is concentrated locally in placer deposits. At sand and gravel plants with often complex extraction, transport and mineral processing technologies, artificially created placer deposits exist at many points in the operations. On account of their usually dark colour caused by the content of iron-rich minerals (magnetite, ilmenite, haematite, etc.), they are easy to detect to the observant eye (Fig. 1). Other common heavy minerals in placer deposits include garnet, zircon, rutile, cassiterite as well as gold and representatives of the platinoids element group.
Hobby gold panners know those places in gravel plants where particularly high concentrations of gold spangles are formed, because of the technological processes, and go about their business there more or less officially. Depending on the yield, the recovered gold is minted into medals, worked into jewellery by goldsmiths or sold in grams to interested buyers. At least in Germany, it can be assumed that the demand for indigenous gold is larger than the gold available. For the year 2017, Germany’s Federal Institute for Geosciences and Natural Resources estimates the gold recovered at German gravel plants at around 25 kg .
Presumably, in most cases, the plant management knows about these activities as a part of the recovered gold is usually given over to the operative company. Apart from a few exceptions, relatively little interest has been shown at sand and gravel plants, in ascertaining whether an official and controlled side-recovery of gold and possibly other heavy minerals is economically viable. The reasons for this attitude are the fact that a high labour requirement is associated with most of the methods applied for the separation of gold from gravel sands, and that quantitative details of the overall procedure are hardly collated nor verifiable.
The simplest data, like, for example, the average gold content in a gravel sand deposit, are systematically and representatively collected only in rare cases. Without this information, it is not known, for instance, how much gold per year passes through the preparation equipment of a gravel plant, and accordingly how much gold could theoretically be recovered. There are still only few published studies on the efficiency of gold separation technology used in industrial conditions at sand and gravel plants (on pilot plant scale e.g. ).
In contrast to the quantitatively and qualitatively precisely known processes in sand and gravel production – starting from the geological exploration of reserves to the end-products – the recovery of gold on the side, especially in their quantitative details, is usually not verifiable and surrounded by a veil of mystery for the operators of surface mines. Most company-external hobby gold panners are not willing to disclose details of their work. The question remains how far these persons systematically determine the above-mentioned basic data themselves in the respective operating plants. For most of them, it is probably a case of what  already wrote under the heading “Several myths among gold panners”, the gist of which is “Of course, I know that I don’t get all the gold out – but no matter, I make my money.”
The studies described in the following were conducted at Kieswerk Meissenheim, a gravel plant owned by the company RMKS Rhein Main Kies und Splitt GmbH & Co. KG, Wesel. Here floating grab dredgers are used to extract material from the over 100-m-thick Quaternary gravel sand deposits of the River Rhine (Fig. 2).
Inspired by the rise in the price of gold from around 2006 and the start of the side recovery of placer gold heralded by great media echo at the Rheinzabern Gravel Plant near Karlsruhe, a stepwise check was performed to ascertain how far the known gold content in the Rhine sediments could have any economic relevance for Meissenheim Gravel Plant, too.
Three other studies of the gold content of the raw gravel sand resulted in a mean value of 2.6 ppb (mg gold per tonne gravel sand). In other words: in one million tonnes of raw gravel sand at the deposit, an average of around 2.6 kg gold is contained.
A representative determination of such a low gold content in the samples by means of a purely chemical process is hardly possible on account of the “nugget effect”. This effect refers to the fact that gold spangles are very irregularly distributed in the natural milieu: depending on the deposition conditions, gold appears to be concentrated in few places; on the other hand, it is depleted or absent at many other places.
The same effect occurs when samples are prepared for chemical analysis. The average mass of the raw gravel sand tested was around 90 kg. This contained an average of 80 gold spangles. For chemical analysis in a laboratory, for practical reasons, the entire 90 kg cannot be chemically broken down. Usually attempts are made to separate a representative subsample from one large original sample (e.g. by means of systematic halving) and only this subsample (in the order of 100 g) undergoes chemical analysis.
If, for example, 90 kg are reduced to 90 g (corresponds to 1/1000), this should contain a representative subsample of 1/1 000 gold spangles (that is 80 spangles / 1 000 = 0.08 spangles). There are, however, only “whole” spangles: subsamples will generally not contain any whole spangles (chemical analysis then results in “0” gold content) and in few cases (approx. in one of twelve subsamples) a whole spangle (chemical analysis then results in an excessively high gold content). Both results are wrong; they deviate widely from the reality.
To mitigate this problem, the gold content in the tested samples was determined purely physically: the entire sample was fed over a 2-m-long launder lined with honeycomb matting, all gold spangles were retained in the honeycombs and picked out manually after further intermediate steps under a binocular. The predominantly very flattened gold spangles were then photographed, and their area measured. Based on multiplication of the area with a mean thickness, the gold volume is calculated and with multiplication with a mean density, the gold mass calculated. Relative to the starting mass of the sample, the gold content was derived. The procedure simplified here is described in greater detail in  and .
Predestined for efficient separation of the gold spangles from the technology chain at the Meissenheim gravel plant seems to be the underflow of the dewatering screen as the gold content increases considerably (24.7 ppb) there as a result of the screening processes and there is a largely continuous stream of material.
The morphological properties of the gold spangles reflect average characteristics that can be encountered in many gravel sand deposits all over the world (Fig. 4). They are characterized by a relatively long transport distance, which on the one hand leads to substantial flattening (formation of plate-like spangles by “hammering”) and associated average spangle thicknesses of around 5 – 15 µm as well as, on the other hand, to mean particle sizes of around 100 µm.
When a suspension with gold-bearing sand is fed over a launder, a large number of physical effects take place, like, for example:
Laminar flow of the suspension in the upper part of the flow body with
• different settling rate of the particles (depending on density, particle size, particle shape) within this laminar zone and
• interaction of the particles (collisions),
• turbulences in the boundary region between the flow body and loops or fibres of the mat with
• different motion of the particles (depending on density, particle size, particle shape) within this turbulent zone as well as
• interaction of the particles (collisions),
• sedimentation of the particles with
• rearrangement of the particles in the upper part of the sediment body and
• sudden and/or continuous vibrations of the plant,
• fluctuating flow rates/quantities,
A physical-numerical modelling of all these processes is not currently possible on account of the complexity of the processes taking place. Accordingly, an optimization of the above-mentioned launder parameters can only be performed empirically by means of local tests.
Such work was conducted at Meissenheim Gravel Plant preferably with a test launder measuring around 7.0 x 0.5 m lined with five mats. The above-mentioned 0 – 1 mm fine sand from the underflow of the sand dewatering screen was fed to the launder as the material stream.
During launder operation, the particles flow and roll over the surface of the mats. The aim is that all gold spangles sediment in the gussets between the individual loops (Fig. 6).
After a certain time, the feed is stopped, the five mats are removed and the material with concentrated heavy minerals trapped in the gussets of each mat flushed out with the help of water sprays. Then for each sample, the entire gold in the 2-m honeycomb mat launder is separated and weighed.
The reason for the test series as explained in more detail below was the question of how long the feed to the gold washing launder should be to obtain optimum operating performance. As mentioned above, an exact modelling of the gold concentration in a gold launder lined with mats “only on paper” cannot currently be realized. In lieu of this, the following assumption is made:
which runs continuously (there are neither empirical documents, nor reason to assume that the saturation of the mat undergoes a sudden change) and
which asymptotically converges towards a limit value (the infinitely long feeding of a launder with gold-bearing sand will not lead to the production of a concentrate of pure gold: at some time, a dynamic equilibrium is achieved in which the same amount of gold is fed at the top end of the launder as is flushed out again at the bottom).
For a very simple mathematical description, a saturation function was used as shown in Fig. 7. If this function is calculated based on determination of as many as possible data pairs (x, y) for a specific gold launder, it is possible, for example, to answer the question what maximum amount of gold can be captured in the launder: this corresponds to the parameter “a” (gold mass or concentration in the mat after an infinitely long feeding duration). The data pairs determined for the test launder are listed in Table 1.
From these data, a regression curve should be drawn up that shows minimal deviation from the measured values – e.g. by means of the method of the smallest error squares. Unfortunately, there no trivial mathematical solution exists for the determination of a and b in the function y = ax / (x + b), when many pairs of x and y exist. For this reason, the variables a and b were determined by means of numerical iteration (“trialling”) in an Excel table as near as possible to the true value (Fig. 8).
The graphical visualization of the cumulative error squares leads to a minimum at a = 1715 and b = 39.5 (Fig. 9).
After an infinitely long feed duration, the test launder will contain a maximum of 1 715 mg gold as well as a gold content of 160 mg/t relative to the entire heavy mineral concentrate retained in the launder. However, an infinitely long feed duration is unrealistic: what therefore is the optimum? The solution of the question results from the slope of the saturation curve: it is highest at the beginning of the feeding to the launder and then decreases steadily (Fig. 11).
If you look, for example, at a total period of 72 hours, one-time cleaning of the mats (clean up) after 72 hours results in 1 107 mg gold (Fig. 12).
Clean-up of the mat every 12 minutes, on the other hand, recovers 3 110 mg gold, but this would have to be done 360 times in the 72 hours.
Manual changing of mats in a launder is a work-intensive procedure. Amongst other things, the feed has to be shut off or diverted. Fastening of the mats has to be loosened; the filled mats are usually heavy and have to be heaved laboriously out of the launder, new mats have to installed and the feed has to be turned on again. This is followed by transport of the mats to a processing facility. There, they are hung up, washed out, the heavy mineral concentrate retained before the gold is further concentrated with the help of different technologies. The cost of this work increases, the shorter the intervals between the cleaning cycles are chosen. Short feed cycles can therefore only be realized with increasingly automated processes.
4 Practical conclusions
After the belt has passed the upper return point, a water jet sprays the underside of the filled loop mat and flushes the heavy mineral concentrate contained in the gussets into a hopper below. As soon as the half revolution is ended (after around 3 minutes), the belt stops, feeding is started up again and a new cycle begins. The concentrate collected in the hopper is then flushed through a bottom discharge to a secondary trough that is lined with honeycomb mats. With a balanced feed of the material stream, water addition and inclination of the launder, as many as possible waste particles (primarily sand, but also unwanted heavy minerals) are flushed away. In the honeycombs remains a concentrate in which the gold content reaches several tenths of a percent.
After the content of the hopper has been emptied and fed over the launder (takes around half an hour), the entire trough is inclined to around 45 ° while additional water is added, as a result of which the gold-rich concentrate collected in the honeycombs is flushed out and collected in a bucket. The processes on the secondary launder are completed after around 40 minutes, so that at the full hour material from the primary launder collected in the hopper can be fed to the secondary launder again (Fig. 13).
The novel feature of the process is that all processes described are performed automatically controlled by a microprocessor; no manual work (changing or flushing of the mats, opening or closing of valves, etc.) is involved. Only in this way can the cleaning cycle be reduced to such a short time span. And a further shortening of the cleaning interval would even be possible, although this does not appear expedient after consideration of all circumstances. The process was protected ; the patent has expired in the meantime.
From a daily fine sand throughput of around 200 t, there results at the end of the shift a bucket of heavy mineral concentrate weighing around 10 kg containing several grams of gold. Further, gold is concentrated to a meltable with the use of a third (tertiary) honeycomb mat launder as well as in downstream procedures. From the separated gold, up to now, two sets of memorial medals have been produced; a third is currently in the process of completion.
The results presented in this paper and the conclusions can contribute to improving the efficiency of the side recovery of placer gold in the sand and gravel industry and placing it on a transparent, quantitatively reliable basis. In addition, a solid basis of fact of the type described promotes the interest of many open mine operators in such a gain based on reliable calculations and plans.
The example of Meissenheim shows that with use of the technologies available today comparatively low gold content can be utilized in an economically viable process. As such mean gold concentrations are present in many other European and worldwide sand and gravel companies, a wide field of development and applications is opened up for the pilot-scale technology for the side recovery of placer gold developed here.
The work presented here were initiated, financed and professionally supported by Dr Gerd Hagenguth, Managing Director at RMKS. Other members of the core team were operations manager Heinz Schlecht as well as Christian Lehmann (now K + S). I thank them and others involved for the harmonious cooperation, which has yielded encouraging results including this publication.
Literatur • Literature
 Elsner, H., Kuhn, K. & Schmitz, M.: Heimische mineralische Rohstoffe – unverzichtbar für Deutschland! Bundesanstalt für Geowissenschaften und Rohstoffe (2017)
 Schiffers, A.: Technische und wirtschaftliche Aspekte der Goldaufbereitung in Kieswerken, Schriftenreihe zur Aufbereitung und Veredlung (2009) 31, Aachen, Shaker
 Clarkson, R.: Gold losses on klondike placer mines – Gold recovery project (Phase I) – Report prepared for the Klondike Placer Miners Association (1989) 61
 LGRB: Karte der mineralischen Rohstoffe 1:50.000 - KMR50 – Kiesmächtigkeiten im Oberrheingraben, abgerufen von http://maps.lgrb-bw.de/?app=lgrb&lang=de am 23.04.2019
 Lehmann, U.: Gold in sächsischen Kiessandlagerstätten, Glückauf (2010) 146, 11, 560-564
 Gutzmer, J., Richter, L., Hennig, S., Petermann, T. & Lehmann, U.: Gold in sächsischen Kies- und Sandlagerstätten, Schriftenreihe des LfULG (2013), Heft 12
 Teschner, B., Smith, N. M., Borrillo-Hutter, T., John, Z. Q. & Wong, T. E.: How efficient are they really? A simple testing method of small-scale gold miners’ gravity separation systems, Minerals Engineering (2017), 105, 44–51
 Taggart, A.F.: Handbook of mineral dressing ores and industrial minerals, John Wiley & Sons, Inc., New York; Chapmann & Hall Ltd., London (1945)
 Patent: Verfahren zur Nebengewinnung von Mineralen hoher Dichte (>3g/cm3) – nachfolgend Schwerminerale genannt –, insbesondere von Seifengold aus im Rohkies enthaltenen Sanden und Schluffen und Anlage, Patent DE102012021317A1 Germany (2014), https://patents.google.com/patent/DE102012021317A1/de