You are on page 1of 16

Simulation and early design studies

Peter Hand, Isandla Coal Consulting David Wiseman, David Wiseman Pty Ltd
Box 3106, Edenvale, 1610, South Africa Box 94, Blackwood, SA 5051, Australia
Phone & Fax: +27 11 455 4755 Phone & Fax: +61 8 8370 2584
petehand@global.co.za david@davidwiseman.com.au

Abstract
Pre-feasibility, feasibility and early design studies call for a particular approach to simulation
and process flowsheeting if such tools are to provide useful assistance to the design process.

Often there is insufficient information available to allow the use of more complex models
needed for detailed simulation and it is necessary to rely on a combination of relatively
simple, physically based models, together with experience of similar processes in operation.

However, there is frequently enough information to allow simulation based sizing of major
equipment more accurate than intuition, particularly in circuits involving recycles. The
information derived from simulation can give broad brush estimates of the number of modules
required, scale of machines needed etc.

Relatively simple simulation can be a useful and important tool in harnessing both formal
design and “rule of thumb” techniques in order to assess basic and advanced plant circuit
design. Oversimplification however is not without its attendant problems including potential
loss of valuable detail.

Examples and case studies are provided of an approach that attempts to walk the line
between sufficient detail to make the simulation worthwhile while working within the bounds of
the quality and amount of data typically available at this stage of the design study.

Keywords

Modelling, Simulation
Introduction
The requirements of pre-feasibility, feasibility and other early design studies present a difficult
environment for process calculations. There is often a lack of detailed process information (in
some cases a lack of any process information). As a consequence, the calculations must be
based on any or all of “standard approaches”, heuristics or broad “rules of thumb”,
“guesstimation”, “what similar operations do”, or “what was used last time”.

In current practice, calculations of this nature are typically performed in a spreadsheet such
as Microsoft’s Excel™ (by far the most common tool of choice for ad-hoc calculations by
process engineers), or by using more specifically mathematically focussed tools such as
Matlab™ or MathCAD™.

In this paper we detail an approach based on using Excel™, but applying structure such that
the implicit ad-hoc nature of the spreadsheet is brought somewhat under control, while still
allowing the user the flexibility of approach that early design study calculations require. The
additional benefits that working in the spreadsheet environment provide are also highlighted.

Document requirements for process design


A typical document set for a process design might include:
• Project Management Documentation
• Process Flow Diagram (PFD)
• Design Criteria
• Piping (or Process) and Instrument Diagram (P&ID)
• Flow Summaries for materials handling equipment sizing.
• Materials balances showing grades and recoveries per process stream.
• Water balances, both in the process and globally for a site.
• Major equipment list
• Reagents and consumables
• Scenario comparisons, trade off studies.
• HAZOP summaries.
Keeping track of these documents, handling change/version control, ensuring that updates in
one document that effect others have indeed flowed through and providing appropriate audit
trails represent a significant problem for groups involved in design studies.

Other issues include the need to document sources of data, laboratory results, calculation
methodology and so on.

The problems are particularly apparent at the early design stage, where the groups involved
may be small, or may in fact be consultants outside of the organisation responsible for later
refinement and implementation of the design. In this instance, there is a further consideration;
handover of the appropriate document set to the next group in the design process.

Not all of the documents listed are appropriate to early design studies, for example, it would
be unusual to develop detailed P&IDs at this point in the design process and process review
stages such as HAZOP studies often depend on those documents.
Application of structure to the spreadsheet
If the documents required for early design can be maintained within a single source
document, many of the above problems can be alleviated. The spreadsheet is a strong
candidate as a suitable host “container” for such documents, since many of them require
calculations in their own right.

Process Flow Diagrams (PFDs)


In smaller companies and increasingly in larger organisations, engineers are performing tasks
that were (prior to personal computing) the province of specialist staff (typists, draftspersons
and the like). While word processing is now a relatively common skill, use of drawing
packages by process engineering personnel for the preparation of process flow diagrams is
far less common, possibly due to the lack of familiarity, perceived level of difficulty and the
learning curve involved with tools such as AutoCAD™, Visio™ and similar software. This then
leads to two typical outcomes – either the user attempts to draw a PFD with familiar but
inappropriate tools (e.g. Word™/Excel™/Powerpoint drawing tools – most often Excel™,
since that is where the data usually resides), or the task is delegated to one of the few
remaining drafting specialists, where it typically becomes a bottleneck in the entire process.

Figure 1. Example Coal Preparation and Handling Plant Flowsheet


The use of drawing/drafting tools that are separate to the calculations also leads to problems
of updating, error checking, mark-up and correction as the iterative design process proceeds.

A possible improvement would be to be to have the ability to work in an environment familiar


to the user and close to the data, such as a spreadsheet, but without the difficulties inherent
in using such a tool. It is this approach which Limn:The Flowsheet Processor aims to
address by providing a “flowsheet aware” drawing environment to allow development of a
PFD within the spreadsheet. Working as an application layer over the spreadsheet, but using
the drawing elements provided by the native spreadsheet drawing tools, Limn allows report
quality PFDs to be prepared easily and quickly.

Figure 2. PFD Showing Calculation Results

Calculation Results on PFD


Through the use of data blocks, which link data tabulated elsewhere in the spreadsheet
directly to the flowsheet for display in proximity to process flow lines or equipment items, or
displayed below the flowsheet in the traditional engineering company PFDs, the results of the
process calculations are automatically updated onto the PFD. This removes the need for the
error prone changing/checking/marking up/correction iteration required when PFDs are
prepared on drafting specific systems.
Structured Modelling Environment
With the availability of the user drawn flowsheet within the spreadsheet, and the ability to
display calculation results directly on that flowsheet, it is reasonable to use the flowsheet to
determine the process connectivity for the calculation directly, removing a further potentially
error prone manual transcription step.

To implement this requires that process unit models are defined in such a manner that the
interconnecting data flows between the model calculations can be rearranged simply. Two
approaches help facilitate this. The first is the definition of a suitable stream structure and the
second is the isolation of the individual unit models from the overall process model.

In the spreadsheet environment, the stream structure can be implemented simply as a range
of cells, one range for each stream. One and two dimensioned arrangements of data are
relatively straightforward, higher dimensioned problems require a little more creativity, but are
still quite achievable.

Stream Summary

Cu
Fine Ore Cu Clnr Cu Rghr Cu Rghr Cu Rghr
Regrind
Feed Tail Conc Feed Tail
Product
tph Galena 4.919 0.55 0.55 1.85 5.47 3.62
tph Sphalerite 16.615 1.05 1.05 1.50 17.66 16.16
tph Tetrahedrite 0.003 0.00 0.00 0.00 0.00 0.00
tph Chalcopyrite 1.707 0.24 0.24 1.39 1.95 0.56
tph FeSG 52.38 1.65 1.65 1.88 54.03 52.15
tph NSG 24.376 1.15 1.15 1.38 25.52 24.14
tph Water 132.558 10.82 10.82 4.31 143.37 225.49

tph Solids 100.00 4.64 4.64 8.00 104.64 96.64

% Galena 4.92 11.87 11.87 23.09 5.23 3.75


% Sphalerite 16.62 22.56 22.56 18.77 16.88 16.72
% Tetrahedrite 0.00 0.04 0.04 0.04 0.00 0.00
% Chalcopyrite 1.71 5.16 5.16 17.35 1.86 0.58
% FeSG 52.38 35.60 35.60 23.48 51.64 53.97
% NSG 24.38 24.77 24.77 17.27 24.39 24.98

% Pb 4.26 10.28 10.28 20.00 4.53 3.25


% Zn 10.80 14.67 14.67 12.20 10.97 10.87
% Cu 0.59 1.79 1.79 6.00 0.64 0.20
% Fe 26.00 19.18 19.18 17.00 25.70 26.42
g/t Ag 114 1,633 1,633 1,650 181 60
% Solids 43.00 30.00 30.00 65.00 42.19 30.00

% Pb Recovery 100.00 11.19 11.19 37.55 111.19 73.64


% Zn Recovery 100.00 6.29 6.29 9.03 106.29 97.26
% Cu Recovery 100.00 14.09 14.09 81.33 114.09 32.76
% Fe Recovery 100.00 3.42 3.42 5.23 103.42 98.19
% Ag Recovery 100.00 66.39 66.39 115.75 166.39 50.63
Figure 3. Typical 1D data structure (flotation by mineral species)
Mass in each SG / Size interval
Feed SG
Floats - 1.30 1.30 - 1.35 1.35 - 1.40 1.40 - 1.45 1.45 - 1.50 1.50 - 1.55 1.55 - 1.60 1.60 - 1.70 1.70 - 1.80 1.80 - 1.90 1.90 - 2.00 2.00 - Sinks
Solids t/h 1,000.00 Size 1.275 1.325 1.375 1.425 1.475 1.525 1.575 1.65 1.75 1.85 1.95 2.125 Water t/h
Water t/h 250.00 +16 28.28 6.325 34.155 36.57 13.225 6.90 5.52 2.53 2.07 0.69 0.345 0.115 6.555 250.00
% Solids 80.0 +8 11.31 13.59 40.921 38.656 14.949 6.644 3.926 2.567 3.473 1.057 0.755 0.604 23.858 Magnetite t/h
Magnetite t/h +4 5.66 16.74 50.406 47.616 18.414 8.184 4.836 3.162 4.278 1.302 0.93 0.744 29.388 0.00
% Ash 25.64 +2 2.83 17.82 36.18 37.44 16.38 6.48 4.32 2.16 2.88 1.44 1.08 1.26 52.56
% Sulphur 0.26 +1 1.41 14.652 29.748 30.784 13.468 5.328 3.552 1.776 2.368 1.184 0.888 1.036 43.216
% Phosphorous 0.13 +0.5 0.71 21.142 9.052 7.254 3.41 1.24 0.992 0.992 0.744 0.682 0.62 0.496 15.376
+0.25 0.35 14.90 7.00 5.85 2.75 1.00 0.80 0.80 0.60 0.55 0.50 0.40 14.85
+0.125 0.18 13.112 6.16 5.148 2.42 0.88 0.704 0.704 0.528 0.484 0.44 0.352 13.068
+0.063 0.09 7.152 3.36 2.808 1.32 0.48 0.384 0.384 0.288 0.264 0.24 0.192 7.128
-0.063 0.04 11.92 5.60 4.68 2.20 0.80 0.64 0.64 0.48 0.44 0.40 0.32 11.88

Figure 4. Example 2D data structure (Size by Density - One stream from a Coal Wizard
generated workbook)

Isolation of individual process unit models requires that inputs to and outputs from the actual
process calculations are defined as separate ranges (mirroring the stream structures), in
much the same way that parameters are passed into and retrieved from a subroutine or
function in a conventional programming language. Once these “interfaces” between the
structured stream data and the process unit calculations are defined and the process unit

Model for Unit: Cyclone

Model Summary

Feed Cyc Overflow Cyc Underflow Audit Check

Solids t/h 369.276 250.00 119.276 OK


Water t/h 451.337 404.624 46.713 OK
w/w % Solids 45.00 38.19 71.858 -
Pulp SG 1.474 1.375 2.055 -
Volume m3/h 556.844 476.052 80.792 -
P80 0.445 0.143 4.643 -
% -0.5 81.172 98.459 46.44 OK

Cyclone Physical Data


Number of cyclones in parallel 5.00
Cyclone Cylinder Diameter - Dc 0.30 m
Equivalent Inlet Diameter - Di 0.10 m
Vortex Finder Diameter - Do 0.15 m
Spigot (Apex) Diameter - Du 0.05 m
Cyclone Cylinder Length - Lc 1.00 m
Cone Full Angle - θ 10.00 degrees
Per Cyclone Volume Throughput - Q 111.369 m3 / hr / cyclone
Volumetric Fraction of solids in feed - Cv 0.189
Feed Pressure at inlet - P 133.133 kPa
Model Constants
d50 constant - KD 5.000E-04
Capacity constant - KQ 300.00
Water split constant - KW 20.00
Sharpness of cut - α 2.10
"fish hook" parameter - β 0.00
Calculated Model Data
d50c (mm) 0.346 mm
Water split to Cyc Overflow - C 0.8965
Water split to Cyc Underflow - Rf 0.1035

Split to Cyc
Size Mean Overflow
+7.5 10.61 0.000
+3.75 5.30 0.000
+1.88 2.66 0.000
+0.94 1.33 0.002
+0.47 0.66 0.103
+0.24 0.34 0.464
+0.12 0.17 0.716
+0.06 0.08 0.819
-0.06 0.04 0.861

Feed Cyc Overflow Cyc Underflow


Water t/h 451.337 404.624 46.713
Size Mean
+7.5 10.61 11.235 0.00 11.235
+3.75 5.30 10.774 0.00 10.774
+1.88 2.66 11.131 0.00 11.131
+0.94 1.33 11.685 0.024 11.661
+0.47 0.66 15.091 1.548 13.543
+0.24 0.34 25.298 11.741 13.557
+0.12 0.17 39.309 28.164 11.145
+0.06 0.08 54.174 44.394 9.779
-0.06 0.04 190.58 164.129 26.45

Figure 5. Example 1D unit model structure (Nageswararao Hydrocyclone) – note process


Feed and Product streams (circled) that are the inputs to and outputs from a “standard”
calculation.
calculations thus modularised, the way is open for simple re-use of spreadsheet based unit
models, both within a particular design study to model multiple similar process equipment
items, or within an organisation, for use in modelling similar processes in different projects.

Automating spreadsheet development


Limn:The Flowsheet Processor carries these basic concepts further again, in the form of
the Limn:WizardPack. Each “Wizard” is a workbook containing pre-defined stream structure
and “template” unit model calculation worksheets, in addition to datablock, summary and
other worksheets that a user may require. When the particular “Wizard” is run, these
worksheets are automatically copied over to the worksheet containing the flowsheet (the
“target workbook”), and the appropriate spreadsheet ranges are associated with the streams
and process unit models depicted on the flowsheet.

The Wizard template workbooks take two forms; those whose pre-defined components are
suitable for a specific process domain (e.g. the Limn:CoalWizard has a two dimensional size
by density structure with unit model worksheets specific to coal processing plants), and those
whose structures and unit models are meant for subsequent user modification and
enhancement (e.g. the Limn:Generic 1D VerticalWizard has a simple one dimensional -
vertical spreadsheet range – stream structure with simple splitter, combiner, reactor type unit
model worksheets).

The template stream data structures and unit model worksheets provided by the
Limn:WizardPack are designed to be further modified by the user, in one of two ways.

The simplest and most common approach, particularly where a process domain specific
wizard is close enough to the final requirements, is to run the Wizard, then to modify the
resulting worksheets in the target workbook, to achieve precisely the stream structure and
unit model nuances desired.

Alternatively, a pre-defined Wizard template may itself be edited and modified to form the
basis of a new “user defined Wizard” containing stream structures and unit models as
explicitly defined by the user, which can then be used in site specific studies or for frequently
required problem domains. The user defined Wizards provide a way of encapsulating process
knowledge held within an organisation into a format that can be used in the future in a
standardised and structured manner, not just by the original Wizard developer, but by others
who perhaps lack the specific process understanding or modelling expertise.

Being provided with pre-packaged data structures and process models in this manner and the
tools with which to ally them to a flowsheet, is attractive to an end user in that it reduces set
up time and speeds up the modelling process. In doing so, it encourages the use of a well
structured approach by default. Adoption of a common approach and consistent structure
within any calculation system has of course, many flow-on benefits, in terms of accessibility
by others, ease of updates, modifications and the general maintainability of the completed
work.
Calculations at appropriate levels of detail
"Everything should be made as simple as possible, but not simpler." A. Einstein
At the early design stage, available process data are usually limited. In these instances, it is
difficult to apply complex models with detailed tuning parameters. The fallback is usually to
adopt a simpler form of model, for example separation efficiency curves, standard breakage
data and so on. In a suitable environment it is possible to combine these with other
established design calculations to size equipment based on the model calculated flowrates.

In the case of crushers, for example, manufacturer’s grading curves may be appropriate, or in
flotation circuits, simple grade (e.g. 2 product formula) or recovery models may suffice to
allow determination of likely process flows, which can then form the basis of equipment sizing
calculations.

Model for Unit: Sec Crusher

Model Summary
Sec
Feed Crusher Audit Check
Product
Solids t/h 327.179 327.179 OK
Water t/h 0.50 0.50 OK
w/w % Solids 99.847 99.847 -
Pulp SG 3.487 3.487 -
Volume m3/h 93.98 93.98 -
P80 118.743 42.005 -
% -25 3.192 43.63 OK

Model Parameters
Crusher Models available NB. This model produces a product size distribution
based on curves fitted to data obtained from various
Hammer Mill manufacturer's literature.
Jaw Crusher (Hard Ore)
The distributions are obviously indicative only, but
Jaw Crusher (Average Ore) should provide a useful starting point where no other
information exists.
Cone Crusher (Hard Ore)

Cone Crusher (Average Ore)

Gyratory Crusher

Coal Dou ble Rolls Crusher

Crusher Closed Side Setting (mm) 32.00

Required
Product Size
Size Mean Distribution
+100 118.74 0.015
+71 84.26 0.398
+50 59.58 3.925
+35 41.83 16.004
+25 29.58 25.815
+18 21.21 20.676
+13 15.30 12.389
+9 10.82 7.296
-9 7.35 13.483

Sec
Feed Crusher
Product
Water t/h 0.50 0.50
Size Mean
+100 118.74 124.89 0.048
+71 84.26 106.998 1.301
+50 59.58 56.649 12.84
+35 41.83 18.864 52.362
+25 29.58 7.009 84.46
+18 21.21 4.292 67.648
+13 15.30 3.043 40.534
+9 10.82 2.236 23.871
-9 7.35 3.198 44.115

Figure 6. Example of a simple crusher model using manufacturer’s “grading curves”.

In Figure 6, a spreadsheet calculation worksheet is shown where manufacturer’s predicted


product sizings are applied to the calculated flowrate into the unit, to estimate a product size
distribution for the chosen crusher closed side setting. The manufacturer’s data, instead of
being read from tables in a product pamphlet, has been digitised and encapsulated in a
spreadsheet macro formula (Excel™ VBA) that can be accessed directly from the
spreadsheet. User input is provided via “option button” user interface elements and entry of
the crusher setting. Errors are reduced, since there is no requirement to look up and
transpose data and the flowrate information is automatically updated and thus correct in terms
of the broader plant model of which the crusher calculation forms a part.

For screens, an efficiency curve approach to determining circuit flowrates may be tied to, for
example, the industry standard VSMA screen sizing calculations. As an example, in figure 7,
the left hand side tables perform the efficiency curve calculations for the simple screening
model. The model cutpoint parameter (d50c) is scaled to the screen aperture which is in turn
derived from deck and panel types chosen by the user (via drop down lists from manufacturer
provided tables stored elsewhere in the worksheet). The calculated flowrate data, together
with the deck and panel information, plus several other required data values are combined on
the lower right section of the page to complete the standard VSMA screen sizing calculations
to produce estimates of bed depth and required screen area.

Model for Unit: Scalping Screen

Model Summary
Scalper Scalper
Feed Audit Check
Oversize Undersize
Solids t/h 500.00 327.179 172.821 OK
Water t/h 5.00 0.50 4.50 OK
w/w % Solids 99.01 99.847 97.462 -
Pulp SG 3.415 3.487 3.291 -
Volume m3/h 147.857 93.98 53.877 -
P80 118.743 118.743 49.628 -
% -25 17.858 3.192 46.013 OK

Equipment Parameters
Screen Slope (degrees from horizontal) 5.00 Screen Sizing Parameters
Screen Width (m) 2.00 Width 2.00 m
Slope 5.00 degrees from horizontal
Model Parameters Velocity 75 ft /2Min = 38.1 cm / sec

Efficiency Curve Parameters - to Scalper Oversize


Application Dry
1
Wet

Alpha 5.00 Feed Bulk Density 1.80 t / m3


Rf 0.10
Efficiency 95.00 Deck Type Heavy1Woven Wire
Aperture Length 50.00 Aperture Description 50 Aperture
7 x 12.5 Wire
Aperture Width 50.00
Nominal Aperture 49.81 Nominal Efficiency 95.00 %
d50c 49.810 Aperture Length 50.00
Aoerture Width 50.00
Partition Fractions
Water split to Scalper Oversize 0.10
Size Mean DECK INFORMATION
+100 118.74 0.999 SYMBOL VALUE FACTOR UNIT
+71 84.26 0.973 500.00 Total Feed TPH
+50 59.58 0.755 U 172.821 = 172.821 UNDER SIZE TPH
+35 41.83 0.377 A 49.81 = 4.51 APERTURE mm
+25 29.58 0.200 B 65.436 = 0.66 OVERSIZE %
+18 21.21 0.143 C 0.178 = 0.40 HALF SIZE %
+13 15.30 0.122 D 1.00 = 1.00 DECK NO. 1,2 or 3
+9 10.82 0.112 E Dry = 1.00 DRY/WET NAME
-9 7.35 0.107 F 1.80 = 1.125 BULK DEN T/M3
G 64.00 = 0.941 OPEN AREA %
H 1.00 = 1.001 OPENING Weibull fit to "H" factor
Feed Scalper Scalper J 95.00 = 1.90 EFFICIENCY %
Water t/h 5.00 0.50 4.50
Size Mean
+100 118.74 125.00 124.89 0.11 DECK AREA CALCULATION
+71 84.26 110.00 106.998 3.002 AREA = 79.262 ft2 = 7.363 m2
+50 59.58 75.00 56.649 18.351
+35 41.83 50.00 18.864 31.136 VALUE FACTOR UNIT
+25 29.58 35.00 7.009 27.991 OVERSIZE : 327.179 tph 327.179 TPH
+18 21.21 30.00 4.292 25.708 BULK DENSITY : 112.39 lb/ft^3 19.575 ft^3/short ton
+13 15.30 25.00 3.043 21.957 VELOCITY : 38.10 cm/sec 75.00 ft / minute
+9 10.82 20.00 2.236 17.764 SCREEN WIDTH : 2.00 m 6.562 ft
-9 7.35 30.00 3.198 26.802 APERTURE SIZE : mm 49.81 mm

Feed DISCHARGE BED DEPTH CALCULATION


Size Mean Distribution BED DEPTH = 2.863 in = 72.721 mm
+100 118.74 25.00 TIMES APERTURE 1.46
+71 84.26 22.00
+50 59.58 15.00 RECOMMENDED DECK SIZE
+35 41.83 10.00 i.e. 2.00 x 3.682 m
+25 29.58 7.00
+18 21.21 6.00
+13 15.30 5.00
+9 10.82 4.00
-9 7.35 6.00

Figure7. Example of a screening model incorporating a standard design methodology with the
process flow calculations.

The above two examples provide specific instances of the generic approach where process
calculations, appropriate to the level of available data, are encapsulated in a unit model
worksheet together with industry standard design heuristics, providing easy and consistent
access to established calculation methods. Using “pre-defined” unit model worksheets in this
way promotes consistency between engineers in the same company, as well as making audit
by outsiders (e.g. consultants providing review functions) much easier .

The encapsulation approach can be extended further to include process unit specific financial
and process calculations on the same worksheet, ensuring that the financial results are based
on up to date and appropriate process data. Reliability is also improved, since the financial
calculations are in close proximity to the process assumptions on which they are based.

Open Architecture – Open Calculations


There is, for issues of speed and security, a temptation to implement aspects of the process
models in higher level language spreadsheet or solution engine callable DLLs. Interestingly
there has been resistance from corporate IT departments who see such proprietary code as
unsupportable, while Excel spreadsheets by virtue of a broader skilled user base are
regarded as less vulnerable to loss of access to the developer. There is increasing corporate
IT pressure for open systems in preference to proprietary code and spreadsheets by their
nature, generally expose their calculations. The various components of Limn assist in this
regard by encouraging use of a structured approach which contributes to spreadsheet
maintainability in the longer term.

Plant Model Solution


While the structured separate process unit model worksheets with defined input and output
spreadsheet ranges as described above could all be linked by normal spreadsheet cell
equations, this can create problems in terms of circular references when the flowsheet
contains recycle. For this reason, and to allow greater flexibility of solution, the Limn package
uses a separate Limn:Solve “solution engine”, implemented in a similar manner to the
Limn:Draw “drawing engine”, as a separate layer over the top of the underlying spreadsheet
worksheets. This “solution engine” handles the data transfers necessary to match the process
topology, schedules the unit models for calculation, checks for convergence and provides
several other useful functions:

Constraint controllers
In many simulation situations, but particularly during early design studies,
where the results are driven by broad requirements (such as a final assay
grade, or a particular flow rate constraint in certain sections of the circuit), it is
necessary to vary certain process parameters to achieve the desired result
from the simulation system. Limn:Solve implements constraint controllers as
part of the flowsheet solution (Figure 8), in a manner analogous to an integral
only plant control loop. This feature adjusts user defined parameters
automatically during the solution to drive the overall system to the desired
values, removing the need for manual adjustments and trial and error
repetitions of the solution.
The constraint controller approach is also extremely useful as a replacement
for “difficult” direct process calculations. In cases where it may be necessary
to calculate the entire system “in reverse” to determine an input value, the
application of a constraint controller to “search” for the solution rather than
performing the calculations explicitly, can be very effective.

Plant Yield
Controlled Variable 38.53
Measured Variable 49.17
Setpoint 49.00
Maximum Output 70.00
Minimum Output 0.00
Controller K -0.001000
Errort-1 -0.173
Control Enabled
Iterations Between Control Actions 1
Start Control at Iteration 1

Lump Fines Split


Controlled Variable 1.04
Measured Variable 29.20
Setpoint 29.05
Maximum Output 10.00
Minimum Output 0.10
Controller K -0.005000
Errort-1 -0.155
Control Enabled
Iterations Between Control Actions 1
Start Control at Iteration 1

Figure 8. Example of constraint control user interface cell ranges during simulation run.

Call back macros for user implemented optimisation code


Before solution, after completion of each iteration, and after solution
convergence, the Limn:Solve solution engine makes calls to standard
named VBA macros. If these are implemented, they run when called,
providing a convenient and consistent interface to any user developed
optimisation methodologies.

Error Checking
The Limn:Solve solution engine must parse the drawn flowsheet to
determine process topology or connectivity. In the process it can optionally
warn the user of flowsheet situations that may not be desirable, such as
isolated (unconnected) process units or streams, missing feed streams and
so on. As the engines accesses the spreadsheet data ranges associated with
units and streams on the flowsheet, if required, it can also perform checks for
consistency and completeness.

Diagnostic and record keeping services


The simulation engine can also provide an output of the process connectivity
(the “connection matrix”) and the flowsheet/spreadsheet range associations
in hard or soft copy format if required.
Sensitivity Analysis
One of the advantages of a simulation system is the ability to quickly run through a series of
“what if” studies to perform sensitivity analyses, or to study the effect of several variables on
overall process operation. To facilitate such studies, the Limn:WizardPack provides a
“Scenarios” worksheet page. This tool allows the user to specify input cell references and
multiple sets of data values to be placed in those cells, prior to each repetition of a simulation.
A corresponding list of cell references, whose values are logged after each simulation run,
provides a record of the simulation results for later analysis.

The input data (independent variables) can be as simple as a set of process cut-points, or as
complex as specifying the blending strategy for a plant fed from multiple mining areas.

Automating the process of testing several/many sets of input data and recording the results
within the same spreadsheet workbook once again means that the relationship between
calculation system and results is maintained, allowing straightforward validation/re-checking
by others, or repetition of the tests, should overall plant or individual process model conditions
change.

4 Seam Washing with Fines Dewatering Screen


Variables Results
Cell Address 1.8000 1.852 74.81 9.16 25.05 5375
Medium Cutpoint
Cell Description Yield Moisture ADCV NCVAR
SG Density
1.30 1.417 22.83 10.78 27.75 5,864
1.32 1.435 27.20 10.13 27.69 5,897
1.34 1.453 31.41 9.68 27.58 5,905
1.36 1.471 35.90 9.31 27.42 5,894
1.38 1.489 40.77 8.99 27.23 5,874
1.40 1.507 45.31 8.76 27.04 5,849
1.42 1.525 49.85 8.58 26.85 5,818
1.44 1.543 54.14 8.42 26.67 5,787
1.46 1.561 80.00
57.60 8.31 26.51 5,759 6,200
1.48 1.579 60.54 8.23 26.36 5,732
1.50 1.596 63.07 8.17 26.23 5,706
1.52 1.614 65.16
70.00 8.13 26.11 5,682 6,100
1.54 1.631 67.02 8.09 25.99 5,658
1.56 1.649 68.61 8.06 25.89 5,636
1.58 1.666 69.91
60.00 8.03 25.80 5,618 6,000
1.60 1.683 71.01 8.01 25.72 5,601
1.62 1.700 71.93 Yield
8.00 25.65 5,585 Yield
1.64 1.717 72.74 7.99 25.58 5,569
1.66 1.734
50.00
73.64 7.97 25.49 5,550
NCVAR 5,900

1.68 1.751 74.73 7.95 25.38 5,526


1.70 1.768 75.78
40.00
7.93 25.27 5,503 5,800
1.72 1.785 76.55 7.92 25.19 5,486
1.74 1.802 77.04 7.91 25.13 5,474
1.76 1.819 77.37 7.91 25.09 5,465
30.00 5,700
1.78 1.835 77.69 7.91 25.04 5,454
1.80 1.852 78.12 7.90 24.97 5,436

20.00 5,600

4 Seam Washing with Fines Dewatering Screen


10.00 5,500

Cyclone Cutpoint Density


0.00 5,400
1.40 1.50 1.60 1.70 1.80 1.90

Figure 9. The Limn:WizardPack Scenarios page showing independent and dependent


variable cell links and data sets with overlaid chart derived from those values.
The spreadsheet as the master design data repository
If we revisit the requirements for a typical document set for the early stages of a process
design in the light of the above discussion concerning the use of appropriate tools to assist in
applying structure and developing consistency in the use of spreadsheets in this role:

(illustrative figures courtesy of Transmin Metallurgical Consultants, Peru)

Project Management Documentation


Standard spreadsheet use.

Process Flow Diagram (PFD)


Achievable in spreadsheet through use of Limn:Draw
Design Criteria
Standard spreadsheet use – can be linked to the inputs to the simulation or to
the setpoints of constraint controllers.

Flow Summaries for materials handling equipment sizing.


Easily derived from stream data summaries in Limn:WizardPack derived
workbooks.
Materials balances showing grades and recoveries per process stream.
Standard output from Limn:Solve flowsheet solution -.can be also displayed
on datablocks on the flowsheet, or in data tables.
Water balances, both in the process and globally for a site.
Easily derived from stream data summaries.

Major equipment list


Standard spreadsheet use – can be linked to the process unit model pages to
allow inclusion of process flow data.

Reagents and consumables


Standard spreadsheet use – process drivers can be derived from standard
limn stream summaries.

Scenario comparisons, trade off studies.


Easily performed using Limn Scenarios tool. Other Excel hosted packages
(@Risk, Crystal Ball) can be used to provide inputs.

The other issues such as the need to document sources of data, laboratory results,
calculation methodology and so on, can of course be accommodated using standard
spreadsheet tools, from simple pasting of scanned documents into appropriate pages of the
workbook (laboratory tests next to the model whose parameters are derived from them etc.),
to more complicated tasks such as importing of assay information about potential feedstock
material from mine geological databases.

Version control is an important issue. Keeping all process calculations and supporting
information in a single source workbook can contribute positively in this area. Having the
process calculations provide the drivers to the other design calculations, and having all
calculations open and available for checking provides a useful audit trail, should it be
necessary to validate or re-do the calculations at a later date.
Using standard tools, it is also possible to provide outsiders with access to the results of
calculations, as well as the supporting documentation, without necessarily giving access to
the detailed calculations. Apart from hard copy printing, printing the workbook pages to an
Adobe™ PDF file provides a convenient medium for electronic transfer.

Conclusions
Existing personal computing desktop tools have sufficient processing power and provide
attractive options as platforms for process design calculations, particularly in the early stages
of a project. The adoption of a standardised, structured approach within a calculation and
design system can leverage the generic power of these environments.

The key to ensuring an efficient and error free approach is the development of the process
unit models and any associated sizing, financial or other calculations in such a manner that
they can be re-used simply and consistently. This can be achieved by careful definition of
process stream structures (the connection between the processes) and associated
spreadsheet ranges and the use of these same structures to define the inputs and outputs for
the individual process unit models.

Automation of the process of combining the user drawn flowsheet with the stream structure,
individual models, summary pages and such other standard worksheets as are required by
the particular application, is also an important step in ensuring consistency of results.

With these “enabling technologies” in place, there is no reason why a spreadsheet cannot
become both the platform for calculation as well as the repository for documentation of the
entire early design study.

You might also like