See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/289245306

A New Approach to Metallurgical Accounting

Conference Paper · October 2012

CITATIONS READS

2 2,621

2 authors:

Stephane Brochot Marie-Véronique Durance

Caspeo Caspeo
37 PUBLICATIONS   75 CITATIONS    24 PUBLICATIONS   110 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

USIM PAC View project

Metallurgical Accounting View project

All content following this page was uploaded by Marie-Véronique Durance on 13 May 2019.

The user has requested enhancement of the downloaded file.

 HOME

A New Approach to Metallurgical

Accounting
S Brochot1 and M-V Durance2

ABSTRACT
In many cases, current metallurgical accounting practice does not consider the errors associated
with the measurements used in obtaining the metal balance. Considering all measurement errors,
including errors associated at each step of the process – from sampling and analytical errors to
measurement errors in circuit instrumentation, can improve the accuracy of the final metal balance
– assisting in the improvement of process performance, optimisation and control.
The consequences of the inclusion of measurement errors on metal balance for different circuit
configurations and typical metallurgical accounting practice (two product and three product
formulae) will be discussed in this paper.
Examples of the tools, methodology of implementation, and periodical metal balance tasks are
presented to illustrate the advantage of this approach in meeting the different requirements of
each department.

INTRODUCTION
If: or a combination of these stages. It is the bridge between
Metal accounting is the estimation of (saleable) metal the technical and the financial point of views of the process.
produced by the mine and carried in subsequent process The process data generated to manage the production
streams over a defined period of time ... (Morrison, 2008a) performances are used to valuate the products and stocks into
financial data.
it has become widely used to quantify the performances of
production plants (metal recovery, losses, environmental The main objective of a metallurgical accounting system is
impact) and to establish an accurate estimation for the metal to help the company in managing process data to generate
inventory (stock taking and work in progress estimation). A a material balance in order to obtain a metal accounting
large discrepancy between the estimated and actual inventory report. The secondary objective is to use the material balance
can have significant financial consequences. Similarly, to accurately calculate the process performances and help
poor estimation of metal recovery and losses can hide the process manager in optimising it. The metal accounting
process issues and give inappropriate production planning. is generally established for a period of production. This
This is why ‘metal accounting provides interface between period can be defined by a regular time period or by the
technical and financial performance measurement’ (Morrison, period of production of a material batch. In accordance with
2008a). These two cultures have two very different points of the financial and accounting rules, the regular time period is
view and have difficulties to conciliate them. The main topic of generally a month.
disagreement is the uncertainty of measurement which implies In the lifetime of a company we can consider three life cycle
uncertainty in the estimation of production and inventories. levels for metal accounting:
The measurement uncertainty and the methods of reducing 1. Metal accounting system life cycle: this begins with the
it have been largely discussed in many papers (Morrison, decision to implement the metal accounting system in a
2008b; Wortly, 2008; Holmes, 2004; 2008; Brochot, 2011). company and finishes with the decision to end it.
The objective of the current paper is to highlight the ways to 2. Production evolution life cycle: this regards the
improve the accuracy of the metallurgical accounting system adjustments of the metal accounting system due to
itself through an implementation process, its granularity (the production evolutions such as a process change, a new
extent of detail to be taken into consideration), and its data production unit, or new products.
reconciliation method and tools. 3. Metal accounting life cycle: this groups the periodical
tasks to obtain a regular metal accounting report.
PROCESS OF METALLURGICAL ACCOUNTING
Metal accounting is a component of the general enterprise Metal accounting system life cycle
accounting. It constitutes a powerful tool to manage metal From the moment a company decides to implement a metal
producing companies at their various stages: mine and mill, accounting system to the time the system reaches completion,
concentrator, smelter or hydrometallurgical plant, refinery, three periods can be identified:

1. Scientific Manager, Caspeo, 3 Avenue Claude Guillemin, BP36009, 45060 Orléans CEDEX 2, France. Email: s.brochot@caspeo.net
2. MAusIMM, Executive Manager, Caspeo, 3 Avenue Claude Guillemin, BP36009, 45060 Orléans CEDEX 2, France. Email: mv.durance@caspeo.net

11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012 217
S BROCHOT AND M-V DURANCE

4. the ‘implementation’ groups all tasks to obtain an reconciliation by material balance, the reporting formats with
operational and efficient metal accounting system KPI calculation rules. All these elements can be managed by
5. the ‘production’ groups all tasks to regularly generate an integrated software solution such as Inventeo (Brochot,
metal accounting reports and update the system according 2011; Caspeo, 2012).
to notable evolutions
6. the ‘closing’ groups all tasks to finalise the last metal Production evolution life cycle
accounting taking into account the plant dismantlement. Many instances may lead to the updating of the metal
The implementation of a metal accounting system is a accounting system: process modification, new raw materials,
company project mobilising all staff: general management, new products, new units of operation, improvements in
financial, accounting, production, laboratory, metrology, the measurement system, improvements or changes in the
information technology, purchasing, sales staff … It necessitates information system, improvements in the material balance
the initiation of a specific project team. Depending on the techniques … these updates necessitate modifying the
initial level of development of the company the following reconciliation model.
tasks have to be taken into consideration: a detailed definition Considering the period from the validation of a reconciliation
of the objectives of the metal accounting and delimitation of model to the validation of an updated version of the
the system; a review of the accounting rules and legislation; a reconciliation model, the following tasks should be executed:
review of the process documentation; a review of the existing the validation of the reconciliation model; the metal account-
measurement system; defining and design implementation of ing calculation and reporting for numerous periods; the
necessary additional measurements; the establishment of the identification of a modification of the metal accounting
measurement uncertainty budget (Brochot, 2011) involving system necessitating an update of the reconciliation model;
the identification and implementation of some improvements; the updating of the reconciliation model; the tests and the
a review of the existing process information system (hardware tuning of the updated reconciliation model in production
and software) with the definition and design of the missing conditions; the validation of the updated reconciliation model
elements; the standardisation of the measurement system; the and normal production with the updated system.
design of the metal accounting system through the material This period must be prolonged to correspond to more than
balance; the configuration of the initial reconciliation model; one metal accounting period. If the reconciliation model has
the tests and tuning of the initial reconciliation model using to be updated for each metal accounting calculation, this is
historical data then in production conditions; and finally an indication that it has not been well configured. A sound
the validation of the metal accounting system and normal compromise between rare updating and the inevitable process
production initiation. evolution is to establish an average of six months between
The period of production starts with the validation of the two updates (no less than three months). A modification in
metal accounting system through the initial reconciliation the metal accounting system must be anticipated. This means
essentially that the metal accounting system has to be taken
model and finishes with the last metal accounting period
into account at the very basis of any modification decision.
before dismantlement. Regardless of the type and frequency
of the material balance, the following tasks must be carried
out: the metal accounting calculation and reporting; the
Metal accounting life cycle
identification and implementation of any improvements in The metal accounting life cycle time commences with the
the measurement system and the data management; changing starting time of the considered period. It finishes with the
metal accounting validation. The validation occurs after the
the system delimitation (system enlargement by taking into
ending time of the considered period, and generally, after
account new production units or system contraction when the
the starting time of the next period. For example, if the metal
units appear out of the scope); all resulting in the updating of
accounting calculation corresponds to one month, the period
the reconciliation model.
starts the first of the month at a given time and finishes
When a decision has been taken to stop the production the first of the next month at the same time. But the metal
and close the site, the following steps are required: the accounting calculation of that period necessitates additional
configuration of the final reconciliation model taking into days waiting missing measurements, intermediate approvals
account the stock exportation, the tailings disposal or the and validations before editing for the final report and final
management of the wastes generated by the dismantlement; validation.
a metal accounting calculation of the last production period; During this time, the following tasks are executed:
the global or periodical metal accounting calculation during • Frequently during the metal accounting period: review the
the dismantlement based on the final reconciliation model; availability of raw data; check raw data using diagnostic
the final reporting; the definitive backup of the databases (for tools; identify all exceptional events which could affect the
possible subsequent control) and the dismantlement of the material balance.
information system. • At the end of the considered period: review the
The ‘reconciliation model’ groups all the information availability of raw data and identify missing data; review
defining the material balance, and the rules and algorithms all exceptional events occurring during the considered
allowing the generation of the metal accounting reports period which can influence the material balance; check
from the raw data generated by the company. This includes raw and basis data using diagnostic tools; verify raw
the definition of the accounting period, the material flow data and fix mistakes; identify the imperative changes
diagram (flow sheet) with the material movements (streams) to introduce into the reconciliation model and update if
and units of operation (nodes), the types of data managed necessary.
in each stream (phase model), the calculation rules to obtain • Waiting missing data: check the availability of missing
basis data resulting from the aggregation of raw data for the data; check raw and basis data using diagnostic tools;
considered period, the calculation rules for the associated verify raw data and fix mistakes; if remaining missing or
measurement errors, the data diagnostic and checking rules, unverified data and mistakes become apparent repeat the
the material conservation constraints, the engine for data previous tasks outlined at a later stage.

218 11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012
 A NEW APPROACH TO METALLURGICAL ACCOUNTING

• When all raw data is available, verified and fixed: complete in the case that the feed mass flow has been measured (the
basis data generation; perform data reconciliation by masses of concentrate and tailings are unknown) and samples
material balance; check reconciled data using diagnostic have been taken and analysed for the three streams, can the
tools and identify the remaining raw data mistakes or ‘two product’ formula be applied to back-calculate the masses
necessary means of adjustment; verify raw data, fix of concentrate and tailings. In this case there is no redundancy
mistakes and return to complete basis data generation; if except in the situation that many contents have been measured
necessary, perform argued adjustment of error and basis for each stream (Guerney, Dunglison and Cameron, 2005).
data and return to data reconciliation; generate provisional The objective of the data reconciliation by material balance
reports; validate the material balance. is to find a set of estimates for the measured values which
• Generation of the metal accounting reports. are as close as possible to the measurements and verify the
Once again, the workflow of these tasks, including material conservation laws. Material balancing allows one
validations and approbations, can be managed by the to describe the material flowing in the process. Knowledge
Inventeo software solution. of the process performances is then improved (Durance,
Brochot and Mugabi, 2004). Sometimes, balancing behaviour
Measurements, measurement error and reveals non-stationary processes or bad accuracy estimation.
basis data The information redundancy allows delivering coherent
The metallurgical accounting is based on the calculation of the estimators more accurate than the initial measurements
material balance of the considered system. This calculation (Ragot et al, 1990; Ragot and Maquin, 2006). This approach
necessitates raw data, such as masses, moisture contents allows for the detection of aberrant values and to reduce error
or assays, which are obtained by measurements. As the due to sampling and measurement.
measurement is a random process, it is subject to uncertainty The objective of a material balance computation is to find a
which can be valuated by its associated ‘measurement error’ set of estimates, which is complete (all streams are perfectly
(Morrison, 2008b). It concerns also the measurements of described), as close as possible of the measured values and in
mass (Wortley, 2008), moisture or metal content, percentage agreement with the material conservation laws.
of solids or density … these last measurements generally
For each stream, the decomposition of the circulating
necessitate sampling, which is the main source of uncertainty
material is as follows:
(Holmes, 2004; 2008). All efforts have to be done to obtain
correct sampling and measurement to avoid any bias. This Qi = quantity of material in stream i; material can be
bias would produce discrepancies between metal accounting dry or wet solids, solution, slurry
and real production with the risk of unacceptable financial Xik = fraction by weight of material class k in stream
consequences. Nevertheless, the variance of the overall i; for metal accounting, the material classes are
measurement error cannot be avoided and its calculation solids and solution, the fractions being the solid and
necessitates the establishment of its uncertainty budget liquid (or moisture) contents
(Brochot, 2011). Tikl = fraction of component l in the material class k in
The quantity of material managed during the considered stream i
period of metal accounting is generally given by the sum of Pil = fraction of component l in stream i; in the case of
many mass measurements such as truck loads or production the slurry, it is the global slurry content
weights per shift. Similarly, the mean moisture or metal
The measured values Qi , Xik , Tikl and Pil never satisfy
contents are calculated by the weighted average of the
constraints of material conservation because of deviation from
contents of many samples. The aggregation of this raw data
the balance position and because of uncertainties associated
gives the ‘basis data’ which is the sum of the total masses
with the measurements. Material balance calculates the
or the mean contents of the material during the accounting
estimates Qi, Xik, Tikl and Pil which firstly satisfy the constraints
period. A measurement error can then be attached to the basis and secondly have the maximum of probability to be the real
data using the error propagation calculation rules (Xiao and values. This last point amounts to seeking the estimates Qi, Xik,
Vien, 2003). Tikl and Pil and which are as close as possible to the measured

Data reconciliation by material balance values Qi , Xik , Tikl and Pil , in accordance with the accuracy
Due to the measurement uncertainty, the basis data is attached to each measurement.
incoherent regarding the material conservation laws Mathematically, this is equivalent to minimise the objective
(Hodouin and Everell, 1980; Herbst, Mehta and Pate, 1988; function:
Fourniguet et al, 1997). To illustrate this incoherence, one
can consider the example of a concentrator with the feed J (Y) = / J (Y) i
as the plant’s input stream and the final concentrate and i
final tailings as the output streams. Assuming there is no by means of the material conservation constraint equations
accumulation of material in the plant (the work in progress (see below), where Y is the vector of estimates (Qi, Xik, Tikl and
is constant) the mass of dry ore feeding the plant during Pil) and where Ji(Y) is the objective function of the stream i (Le
the period has to be equal to the sum of the masses of dry Guirriec, Brochot and Bergounioux, 1995; Le Guirriec, 1996).
concentrate and dry tailings produced during the same Ji(Y) can be expressed as:
period. If dry mass measurements have been performed for 2

e o / c Xv -XX m / c Pv -PP m
2 2
these three streams, their values have little chance in verifying Qi - Qi ik ik il il
this conservation law. Similarly, the masses of the metals are Ji (Y) = + +
vi Q i k ik ik 1 il il
incoherent due to the uncertainty of the dry mass and metal
2
content measurements. The incoherence can be observed
when there is data redundancy: when there is more data than + / c Tv -TTikl

ikl ikl
ikl
m
the required minimum to calculate the material balance. Only k,l

11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012 219
S BROCHOT AND M-V DURANCE

Minimising the objective function is equivalent to The BILCO software was developed by BRGM to solve
maximising the function: material balance problems in the field of mineral processing,
-Ji(Y) which is characterised by its great complexity in the material
e
-J(Y)
= % e description. It offers an interactive, quick and accurate way
i of solving material balance problems in a vast number of
This function is the product of the normal distributions of applications. It has been used for metal accounting in association
probability of the measurements. The maximum probability is with spreadsheets using its component object modelling (COM)
then reached and the estimates are the more probable values interface (Durance, Brochot and Mugabi, 2004; Muza, 2005).
for the material balance. They are not the true values which Now, the Inventeo solution offers a complete range of tools
will remain unknown, by definition of the random nature for metal accounting directly addressing databases without
of the measurements. It is then assumed that the random spreadsheets as recommended by the AMIRA code (AMIRA,
variables associated with the measurements are following a 2007; Morrison and Gaylard, 2008).
normal distribution. If this assumption can be drawn for raw
data, it is less likely the case when basis data is calculated
from many raw data (Xiao and Vien, 2003). GRANULARITY OF MATERIAL BALANCE
This objective function uses the probability distribution of The quantity of information used to calculate the material
the measurements of the metal content, Tikl for example, not balance has a large impact on the accuracy of the metal
the partial quantity of that metal, QiXikTikl, as it is commonly accounting. Lyman (2005) highlighted the improvement
the case (Morrison, Gu and McCallum, 2002). This is in the material balance accuracy when using intermediate
particularly advantageous when the content measurements process streams. The extent of detail, the granularity, of the
are more accurate than the mass measurements. system concerns also the accounting frequency, the material
The constraints can be divided into material conservation flow diagram, the material characteristics by using multi-
constraints, user-defined constraints and data integrity phase and multi-element balancing, the way to estimate the
constraints. The material conservation constraints are attached inventory of stocks and work in progress (WIP), or the raw
to each unit of equipment. The elements of the incidence matrix data used to calculate the basis data.
Mij have the value +1 when the stream i is an input stream of
the node j, -1 when the stream i is an output stream of the Accounting frequency
node j, and 0 when the stream i is not linked to the node j. The The metal accounting is generally established for a period
basic conservation laws are: of production. This period can be defined by a regular time
The global quantity conservation around node j: period or by the period of production of a material batch.
The regular time period is generally a month from a fixing
/M Q = 0 ij i time of the first day of the month to the same time of the first
i day of the next month. It can be also a quarter or a year for a
The partial quantity conservation for material class k around quarterly or yearly accounting, a day or a shift for the process
node j: follow-up. The batch approach is mainly used for process
follow-up but can be included and consolidated into a time
/M Q X ij i ik =0 period balance for metal accounting. In both cases, the metal
i accounting is an iterative task which is repeated regularly.
The partial quantity conservation for component l around In most companies, the financial department imposes a
node j: monthly metal accounting. It is generally associated to a
monthly stock taking allowing a more accurate estimate
/M Q P ij i il =0 of the stocks and WIP (see below). Quarterly and yearly
i
metal accountings are generally obtained by summing the
The partial quantity conservation for component l into individual monthly accountings. If this method is more or less
material class k around node j: imposed by accounting rules and legislation, it is not the more
accurate. Indeed, calculation of the material balance over
/M Q X ij i ik Tikl =0
a longer period has the advantage to reduce the beginning
i
(opening) and ending (closing) inventories compared to the
It is possible to sum some conservation constraints around total material movements. As it is usually admitted that the
node j to obtain a new conservation constraint which can be
stock estimates are less accurate than the material movements
used to take into account, for example, the phase transfer in a
(Wortley, 2008; Holmes, 2008; Connelly, 2009), the weight of
leaching or precipitation model.
the variance of the stock estimates into the global variance
The data integrity constraints are the closure constraints, of the metal accounting is reduced. In addition, specific
such as the 100 per cent fitting and the mean ratio constraint: stock taking can be carried out annually with more accurate
methods of measurement.
1- /X ik =0 1- /P il =0 1- /T ikl =0
Conversely, shorter period accounting is generally less
k l l
accurate. If it is interesting to calculate material balance
Pil - /X ik Tikl =0 per day or per shift for process follow up and production
k reporting, it cannot be used as it is for metal accounting.
The resolution of the system is based upon the Lagrange Indeed, there is no stock-taking at this frequency and all
formalism and the iterative algorithms Newton-Raphson and inventories are just rough estimates with a large associated
Ito-Kunish (Le Guirriec, 1996). After resolution, the variance- error (Cargill, Freeman and Gilbertson, 2002; Muza, 2005).
covariance matrix of the estimates can be deducted and the The intermediate frequency can be used, such as on a weekly
error associated with the estimates can be calculated. It shows basis in gold mines (Guerney, Dunglison and Cameron, 2005;
the accuracy improvement due to data redundancy. Guerney, 2008).

220 11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012
 A NEW APPROACH TO METALLURGICAL ACCOUNTING

Material flow diagram /M Q X ij i is =0

Most of the metallurgical accounting systems reduce the i
material flow diagram to one node representing the plant The advantage of these formulas is to keep the individual
and to the input and output streams (Morrison, Gu and accuracy of wet mass and solid content measurements. It
McCallum, 2002; Jansen, Morrison and Dunn, 2007). In the also increases the redundancy when the water balance can be
case of a mining company, reconciliation between the mine simultaneously calculated.
and the concentrator through the intermediate stockpiles and/
When considering hydrometallurgical plants, the multi-
or blending beds is considered (Morrison, Gu and McCallum,
2002; Morley and Moller; 2005). The one node approach is phase balance is absolutely necessary (Deglon and Gaylard,
also used per plant area. In this case, the material balance is 2008). Considering an ore leaching plant or selective precip-
calculated sequentially with the risk to accumulate errors from itation plant, the handled material is slurry with solids (ore
one area to the next (Cargill, Freeman and Gilbertson, 2002). or precipitates) and solution. Both are characterised by their
metal contents (metals or other elements if they are relevant
The redundancy of data is a way to improve the accuracy
for metal accounting). The variables are then:
of the material balance. Due to the propagation of variance
through the material conservation laws, the error associated Qi = mass of slurry in stream i
to the reconciled data (estimates resulting from the data Xis = fraction by weight of solids in stream i
reconciliation by statistically coherent material balance) is
Xil = fraction by weight of liquid (solution) in stream i
lower than the error associated to the corresponding basis data
derived from measurements. A way to add redundancy to the Tism = fraction of metal m in the solids in stream i
system is to use a more detailed flow sheet with intermediate Tilm = fraction of metal m in the liquid in stream i
streams and nodes. It has to be limited to the streams for which
with the constraint:
there are measurements: weighing, sampling, assaying. The
number of streams without measurement has to be reduced
to its minimum. Some units of operation have to be grouped 1 - Xis - Xil = 0
in one node if they have ‘unknown’ streams between them.
For example, in a closed grinding circuit, if the circulating The material conservation laws are the conservation of
load is unknown (neither weightometer nor sampling), the slurry which is always used (except if some losses, such as
mill and the hydrocyclones are grouped in one node with the evaporation, cannot be measured):
circuit feed as the input stream and the overflow as the output
stream. /M Q = 0 ij i
i
When using only one node (the plant) for the material
balance, the inventories (stocks and WIP) are globalised for the conservation of solids which is used for all process stages,
that node and accounted all together. When using intermediate for which the solids are neither dissolved nor precipitated
nodes, each inventory can be individually accounted through such as decantation, filtering, classification, or cyanidation
a specific node for the stocks or a unit operation node for (for which the loss of ore mass is negligible compared to the
the WIP. Stock variations are then directly linked to their measurement errors):
input and output streams. This adds redundancy and then
improves the accuracy of the inventory. /M Q X ij i is =0
i
The accuracy improvement when increasing the granularity the conservation of liquid which is used for all process stages
of the flow diagram has been theoretically proved and in which the solids are neither dissolved nor precipitated:
illustrated in particular cases (Lyman, 2005).
/M Q X ij i il =0
Multi-phase and multi-element balancing i

It is always preferable to use raw data directly derived the conservation of metal m in the solids which is used for
from a measurement than results of calculation from many all process stages in which the metal is neither dissolved nor
measurements. It is the case with the wet mass and the precipitated:
moisture content. Most of the metal accounting systems use
the dry mass for solid material balance (Morrison, Gu and /M Q X ij i is Tism =0
McCallum, 2002; Guerney, Dunglison and Cameron, 2005; i

Jansen, Morrison and Dunn, 2007). The dry mass is well the conservation of metal m in the liquid which is used for all
derived from the measured wet mass and moisture content the process stages in which the metal is neither dissolved nor
but only the dry mass is used for data reconciliation. If the precipitated:
measurement system allows the calculation of the water
balance, it is preferable to use the wet mass and the solid /M Q X T ij i il ilm =0
i
content (complement of the moisture content) for the solid
balance. The variables are then: In the case of a phase change (solid dissolution or
precipitation), the global mass has to be considered for the
Qi = mass of material (solids and water) in stream i
metal m:
Xis = fraction by weight of solids in stream i
and the material conservation laws are the conservation of /M Q X T +/M Q X T
ij i is ism ij i il ilm =0
material (solids and water: wet solids, slurry): i i

All these equations are mass based while the measurements

/M Q = 0
ij i are generally volume based (tank volume, volumetric
i flowmeters). The slurry density has to be measured and the
and the conservation of dry solids: mass error calculation is based on its measurement error.

11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012 221
S BROCHOT AND M-V DURANCE

The simultaneous material balance calculation of many INVENTEO SOLUTION

components adds redundancy to the system and improves The Inventeo solution (Caspeo, 2012) offers a set of tools
the accuracy of the metal accounting (Morrison, Gun and capable of addressing most of the requirements described
McCallum, 2002; Guerney, Dunglison and Cameron, 2005). above through the reconciliation modeller. The metal
The choice of the components to balance is crucial for the accounting frequency can be configured and various scenarios
quality and the accuracy of the material balance. The metals (or can be considered depending on the availability and quality
components) of economical interest have to be considered. But of raw data. The material flow diagram can be drawn with
other components can be accounted for their ability to improve the desired level of details. The phase model tool permits the
the material balance. That is the case of major components for description of the material in terms of many phases with up
which accurate content can be measured. These components to two levels of description (such as solid/liquid contents
and metal contents). Each basis data can be obtained by
can enhance the poor estimates of some masses. Some
aggregation of raw data coming from third party databases.
components which are clearly separated in some units of Similarly, variance models calculate the measurement error
operation can improve the material balance whereas some associated to each basis data. The material conservation laws
masses cannot be measured. In pyrometallurgical plants, can be easily generated and modified to take into account
where the mass conservation cannot be verified due to gas particular cases which can occur. The analysis tools, based
losses, some components report only in the measured streams on check in, check out methods, help in identifying abnormal
such as Si in a furnace reporting to matte and slag. basis data outside the acceptable incoherence due to their
level of confidence.
Inventory of stocks and work in progress The data reconciliation engine calculates the statistically
There are two methods to manage the inventories: coherent material balance in one pass. All the streams and
1. The quantity and quality of the real stock are accounted components are simultaneously estimated taking into
for at the beginning (opening) and at the end (closing) of account the level of confidence (accuracy) of each basis data.
the considered period. The beginning inventory is then At the end of the material balance calculation, the errors
considered as an input stream and the ending inventory associated to the estimated values are then calculated. These
errors, compared to the error of the corresponding basis
as an output stream, the ending inventory becoming the
data, quantify the improvement of the metal accounting. The
beginning inventory of the following period.
analysis tools, based on statistical comparison between basis
2. Only the variation of the stock is considered as an output and reconciled data, are able to point out any issues such as
stream with a positive mass value when the stock is bias or unaccounted losses.
growing and a negative value when it is decreasing.
Naturally, the first method is preferable as it is closer to the CONCLUSIONS
reality and the stock taking gives an estimate of its quantity
A metallurgical accounting system has to conciliate two
and quality at a given time. The second method can be used in
points of view: the technical point of view for which the
case of uncertain stocked quantity which cannot be measured
material balance is the product of a statistical approach of
directly. the reality and the financial point of view for which the metal
As previously pointed out, the stock taking is difficult to balance refers to an exact and coherent economical value in
perform and less accurate than the material movements the accounting system. Nevertheless, the material balance is
(Wortley, 2008; Holmes, 2008; Connelly, 2009). The WIP based on measurements which are random processes. The
measurement is difficult and generally much more inaccurate. measurement error has to be considered when corrections
It is sometimes preferable to ‘calculate’ the inventory (mass take place during the data reconciliation process. If the data
and contents) from data which have been measured in reconciliation is based on a statistical coherent material
material flow conditions. In that case, a ‘stock model’ is balance, as proposed in this paper and implemented in the
Inventeo solution, the obtained estimated values are the more
used, considering the feeding streams and the reclaiming
probable ones. These values can be accepted by the financial
streams with their timetable (quantity and quality of material point of view if they have not to be adjusted in the future due
versus time). The stock can be considered as a perfect mixer to wrong estimation. Any possible future adjustment has to be
(such as an agitated tank), a FIFO (first input, first output) as small as possible and taken into account as soon as possible.
or a LIFO (last input, first output). Using this method, it is It can be then accounted as gains and losses reported with
preferable to identify each stock (stockpile, tank, etc) with a known and unknown losses. Only an accurate, coherent and
node in the material flow diagram. Indeed, the main risk is to transparent metal accounting system can offer this advantage.
accumulate the discrepancies between the expected and real Many ways for the improvement of the accuracy of the
production into the stocks and, after many months, obtain a metallurgical accounting system have been reported in this
figure far from the reality. The management of small stocks article. The implementation of such a system appears as a
with frequent depletion can reduce this risk and yields a more company key project requiring continuous improvements
accurate material balance. linked to the process changes and measurement system
More advanced stock modelling may be required when enhancement. If a solution such as Inventeo can address most
stock taking is too difficult or costly, such as heap leaching of these requirements, the implication of the personnel of the
(Gebhardt and Cross, 2008) or cathodes in progress which can company constitutes the main key to its success.
be accurately obtained by indirect measurement.
REFERENCES
Basis data calculation AMIRA P754, 2007. Metal accounting, code of practice and guidelines:
As defined above, the basis data is the result of the sum of Release 3.
the masses or weighted average of contents. It is preferable Brochot, S, 2011. The application of sampling theory in metallurgical
to use raw data than data processed by another system (such accounting process – Inventeo methodology implementation,
as historian). The main difficulty resides in the association in Proceedings Fifth World Conference on Sampling and Blending
between masses and contents in order to calculate a weighted (eds: M Alfaro, E Magri and F Pitard) pp 185-193 (Gecamin:
average which is more accurate than a simple mean. Santiago de Chile).

222 11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012
 A NEW APPROACH TO METALLURGICAL ACCOUNTING

Cargill, R, Freeman, N and Gilbertson, R, 2002. Metal balance for Jansen, W M, Morrison, R D and Dunn, R, 2007. Metallurgical
Olympic Dam using a standard industrial software package, accounting in the Northparkes concentrator – A case study, in
in Proceedings Value Tracking Symposium 2002, pp 149-156 (The Proceedings Ninth Mill Operators’ Conference, pp 157-167 (The
Australasian Institute of Mining and Metallurgy: Melbourne). Australasian Institute of Mining and Metallurgy: Melbourne).
Caspeo, 2012. Inventeo – User’s manual, 156 p (Caspeo: Orleans). Le Guirriec, E, 1996. Modelling and mathematical solution to a
problem of statistically coherent material balance (in French),
Connelly, D E G, 2009. Metallurgical accounting standards in
PhD thesis, University of Orléans, France.
process plants, in Proceedings Tenth Mill Operators’ Conference,
pp 281-287 (The Australasian Institute of Mining and Metallurgy: Le Guirriec, E, Brochot, S and Bergounioux, M, 1995. An augmented
Melbourne). lagrangian method for problems arising in mineral processing,
in Proceedings 17th IFIP TC7 Conference on System Modelling and
Deglon, D and Gaylard, P, 2008. Hydrometallurgical refineries, Optimization (eds: J Dolezal and J Fidler) pp 1:65-68.
in An Introduction to Metal Balancing and Reconciliation
(ed: R D Morrison), pp 383-407 (Julius Kruttschmitt Mineral Lyman, G J, 2005. Sampling, metallurgical accounting and reduction
Research Centre: Indooroopilly). of balance estimation variance, in Proceedings Second World
Conference on Sampling and Blending, pp 127-135 (The Australasian
Durance, M V, Brochot, S and Mugabi, M, 2004. Material balance Institute of Mining and Metallurgy: Melbourne).
approach for parameter determination in Bioleaching process, in
Morley, C and Moller, R, 2005. Iron ore mine reconciliation – A case
Proceedings 9th International Symposium on Computer Applications
study from Sishen iron ore mine, South Africa, in Proceedings Iron
in Biotechnology (CAB9), pp 121-126, Nancy.
Ore Conference 2002, pp 311-318 (The Australasian Institute of
Fourniguet, G, Villeneuve, J, Rocchia, L, Védrine, H, Brochot, Mining and Metallurgy: Melbourne).
S and Guillaneau, J C, 1997. Material balance analysis of a
Morrison, R D, 2008a. Motivation for and benefits of accurate metal
hydrometallurgical process for the recylcing of electric furnace
accounting, in An Introduction to Metal Balancing and Reconciliation
wastes (in French), in Proceedings Sixth French Congress of Process (ed: R D Morrison), pp 3-23 (Julius Kruttschmitt Mineral Research
Engineering (Société Française de Génie des Procédés: Paris). Centre: Indooroopilly).
Gebhardt, J and Cross, M, 2008. Heap leach simulation and software Morrison, R D, 2008b. Basic statistical concepts for measurement and
tools, in An Introduction to Metal Balancing and Reconciliation sampling, in An Introduction to Metal Balancing and Reconciliation
(ed: R D Morrison), pp 475-494 (Julius Kruttschmitt Mineral (ed: R D Morrison), pp 27-75 (Julius Kruttschmitt Mineral
Research Centre: Indooroopilly). Research Centre: Indooroopilly).
Guerney, P, 2008. Gold leach plants, in An Introduction to Metal Morrison, R D and Gaylard, P G, 2008. Applying the AMIRA P754
Balancing and Reconciliation (ed: R D Morrison), pp 361-382 (Julius code of practice for metal accounting, in Proceedings MetPlant 2008,
Kruttschmitt Mineral Research Centre: Indooroopilly). pp 3-22 (The Australasian Institute of Mining and Metallurgy:
Guerney, P, Dunglison, M and Cameron, P M, 2005. Leveraging Melbourne).
accuracy and precision – Multi-phase mass balancing and Morrison, R D, Gu, Y and McCallum, W, 2002. Metal balancing from
reconciliation as a tool for quality data management, in Proceedings concentrator to multiple ore sources, in Proceedings Value Tracking
Second World Conference on Sampling and Blending, pp 121-125 (The Symposium 2002, pp 141-147 (The Australasian Institute of Mining
Australasian Institute of Mining and Metallurgy: Melbourne). and Metallurgy: Melbourne).
Herbst, J A, Mehta, R K and Pate, W T, 1988. A hierarchical procedure Muza, P, 2005. Process information management system at Mount
for mass balance closure and parameter estimation: Application Isa, in Proceedings First Extractive Metallurgy Operators’ Conference,
to ball mill grinding, in Proceedings XVI International Mineral pp 179-187 (The Australasian Institute of Mining and Metallurgy:
Processing Congress (ed: K S E Forssberg), pp 1719-1732 (Elsevier: Melbourne).
Amsterdam). Ragot, J, Darouach, M, Maquin, D and Bloch, G, 1990. Data validation
Hodouin, D and Everell, M D, 1980. A hierarchical procedure for and diagnosis (in French), (Hermes: Paris).
adjustment and material balancing of mineral process data, Ragot, J and Maquin, D, 2006. Data validation and reconciliation.
International Journal of Mineral Processing, 7:91-116. conventional approach difficulties and developments, Les
Holmes, R J, 2004. Correct sampling and measurement – The Techniques de l’Industrie Minérale, 29:22-30.
foundation of accurate metallurgical accounting, Chemometrics Wortley, M, 2008. Mass measurement, in An Introduction to Metal
and Intelligent Laboratory Systems, 74:71-83. Balancing and Reconciliation (ed: R D Morrison), pp 77-139 (Julius
Holmes, R J, 2008. Sampling, in An Introduction to Metal Balancing and Kruttschmitt Mineral Research Centre: Indooroopilly).
Reconciliation (ed: R D Morrison), pp 141-170 (Julius Kruttschmitt Xiao, Z and Vien, A, 2003. Limitations of variance analysis using
Mineral Research Centre: Indooroopilly). propagation of variance, Minerals Engineering, 16:455-462.

11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012 223

View publication stats