CA3095875C - Soft sensing of chemical variates of a process stream from a bitumen extraction operation - Google Patents

Abstract

A method for producing bitumen comprising producing multiple process streams from a bitumen-containing aqueous slurry, the multiple process streams comprising a target stream and a measured stream; measuring real-time values of at least two physical variates of the measured stream; and determining a real-time value of a predicted chemical variate of the target stream by correlating the at least two physical variates, such correlation being derived from a soft sensor. Each of the multiple process streams can be a paraffinic froth treatment (PFT) process stream. The target stream can be the same as the measured stream. The at least two physical variates that are directly measured can be a combination of pH, temperature, vapour pressure, density, viscosity, or flow rate. The chemical variate that is predicted via use of the soft sensor can be bitumen content, water content, solvent content, solids content, asphaltenes content, residual metals content, naphthenic acids content, total acid number (TAN), calcium content, clay content, or fines content. The method can further include executing a corrective action when the real-time value of the predicted chemical variate deviates from the threshold value.

Description

SOFT SENSING OF CHEMICAL VARIATES OF A PROCESS STREAM FROM A
BITUMEN EXTRACTION OPERATION
TECHNICAL FIELD
[0001] The technical field generally relates to bitumen extraction operations, and more particularly to techniques for predicting chemical variates of a process stream based on a set of measured physical variates.
BACKGROUND

[0002] In primary bitumen extraction operations, mined oil sands ore is crushed and mixed with water to form an oil sands slurry that is pumpable for further bitumen extraction operations. The oil sands slurry is then separated, via gravity-based separation in a primary separation vessel, into multiple streams including a bitumen froth and solids-enriched tailings.

[0003] In subsequent secondary bitumen extraction operations, the bitumen froth can be diluted with a diluent to assist in the removal of remaining minerals and water.
When the diluent is a paraffinic solvent, the bitumen froth treatment operations can be referred to as paraffinic froth treatment (PFT) operations. The bitumen froth is separated into diluted bitumen and a solvent diluted tailings component in a froth separation unit (FSU), which can include two or three settlers arranged in a counter-current configuration. The diluted bitumen can then be supplied to a solvent recovery unit (SRU) to produce recovered solvent and solvent recovered bitumen, while the solvent diluted tailings component can be supplied to a tailings solvent recovery unit (TSRU) to produce recovered solvent and solvent recovered tailings, which can also be referred to as froth treatment tailings. The solvent recovered tailings can be further processed or can be supplied to a tailings disposal site for settling.

[0004] The solvent recovered bitumen, which can be referred to as a bitumen product, is then sent to an upgrader or diluted to be shipped to refineries, for further conversion into crude oils.

[0005] Multiple physical and chemical parameters, including the composition of the treated bitumen streams, can affect the performance of each step of primary and Date Recue/Date Received 2020-10-09 secondary extraction operations as well as the quality of the final bitumen product. For example, high clay contents in oil sands ore can be responsible for lower bitumen recovery rates. Clay can indeed be an undesirable component of bitumen streams and can interfere with bitumen separation mechanisms in both primary and secondary extraction operations. Processing oil sands ore having a high clay content can lead to reduced bitumen recovery, lower efficiency in terms of primary separation vessel (PSV) performance, and poorer bitumen product quality.

[0006] In the context of PFT operations, there are challenges related to monitoring various bitumen streams and implementing process control strategies, for reducing the likelihood of off-specification streams and ensuring quality of the bitumen product.
SUMMARY

[0007] Soft sensors can be used instead of more complicated systems, such as NIR
analyzers, to obtain real-time values of sensed chemical variates of a process stream deriving from bitumen extraction operation, such as PFT process streams. Soft sensors as described herein can avoid relying on sample collection and laboratory analysis, and can thereby reduce delays for recovering relevant information on multiple process streams. Soft sensors offer reliable online measurement of chemical variates, facilitating the adjustment of operational conditions of the bitumen froth treatment operation in real-time, thereby enhancing control of the bitumen product quality.

[0008] In one aspect, there is provided a method for producing bitumen comprising:
(i) producing multiple process streams from a bitumen-containing aqueous slurry, the multiple process streams comprising a target stream and a measured stream;
(ii) measuring real-time values of at least two physical variates of the measured stream, wherein the at least two physical variates are a combination of pH, temperature, vapour pressure, density, viscosity, or flow rate; and (iii) determining a real-time value of a predicted chemical variate of the target stream by correlating the at least two physical variates, wherein the predicted chemical variate is bitumen content, water content, solvent content, solids Date Recue/Date Received 2020-10-09 content, asphaltenes content, residual metals content, naphthenic acids content, total acid number (TAN), calcium content, clay content, or fines content.

[0009] In a typical bitumen froth treatment installation, producing the multiple process streams from the bitumen-containing aqueous slurry can include producing a bitumen froth stream from a Primary Separation Vessel (PSV), a diluted bitumen froth stream being fed to a Froth Separation Unit (FSU), a diluted bitumen overflow stream from the FSU, a bitumen product stream from a Solvent Recovery Unit (SRU), a recovered solvent stream from the SRU, a tailings stream from a Tailings Solvent Recovery Unit (TSRU), or a combination thereof. In an implementation where a paraffinic solvent is added to a bitumen froth stream separated from the bitumen-containing aqueous slurry, each of the multiple process streams can be a paraffinic froth treatment (PFT) process stream.

[0010] In some implementations of the method, the predicted chemical variate can be:
asphaltenes content and the at least two physical variates comprise temperature, density and flow rate;
water content and the at least two physical variates comprise temperature, density and flow rate;
solids content and the at least two physical variates comprise temperature, density and flow rate;
asphaltenes content, water content, solids content or residual metals content and the at least two physical variates comprise temperature, flow rate and viscosity;
solvent content and the at least two physical variates comprise temperature, vapour pressure, and flow rate.

[0011] For example, the measured stream can be the same as the target stream.
In this implementation, the target stream can be the diluted bitumen overflow stream, the predicted chemical variate can be an asphaltenes content of the diluted bitumen overflow stream, and the at least two physical variates can include temperature, density and flow rate. The target stream can further be the diluted bitumen overflow stream, the predicted chemical variate can be a water content of the diluted bitumen overflow Date Recue/Date Received 2022-09-29 stream, and the at least two physical variates can include temperature, density and flow rate. The target stream can further be the diluted bitumen overflow stream, the predicted chemical variate can be a solids content of the diluted bitumen overflow stream, and the at least two physical variates can include temperature, density and flow rate.
The target stream can further be the bitumen product stream, the predicted chemical variate can be an asphaltenes, water, solids or residual metals content of the bitumen product stream, and the at least two physical variates can include temperature, flow rate and viscosity.

[0012] In another example, the target stream can be different from the measured stream. In this implementation, the target stream can be the solvent recovery stream, the measured stream can be the bitumen product stream, the predicted chemical variate can be a solvent content of the solvent recovery stream, and the at least two physical variates can include temperature, vapour pressure, and flow rate. The target stream can further be the bitumen product stream, the measured stream can be the bitumen-containing aqueous slurry, the predicted chemical variate can be the TAN, naphthenic acid content or calcium content, and the at least two physical variates can include pH
and temperature.

[0013] In some implementations of the method, the predicted chemical variate and the at least two physical variates can be correlated using a multivariate model comprising multiple equations correlating multiple chemical variates to multiple physical variates, each physical variate being attributed a weighting in each equation of the multivariate model. Optionally, the method can further rinclude determining a further dependent chemical variate based on the predicted chemical variate. For example, the predicted chemical variate can be asphaltenes content and the further dependent chemical variate can be nickel and vanadium content.

[0014] In some implementations of the method, the real-time values of the at least two physical variates are detected using at least one in-line sensor.

[0015] The method can include measuring multiples physical variates and predicting multiple chemical variates. In some implementations of the method, the target stream can be a first target stream, the measured stream can be a first measured stream, the multiple process streams can further include a second target stream and a second measured stream, and the method can further include:

Date Recue/Date Received 2020-10-09 (iv) measuring the real-time value of each of the at least two physical variates of the second measured stream; and (v) determining the real-time value of the predicted chemical variate of the second target stream by correlating the at least two physical variates; and (vi) comparing the real-time value of the predicted chemical variate of the first target stream determined via step (iii)with the real-time value of the predicted chemical variate of the second target stream determined via step (v), to assess the efficiency of a bitumen froth treatment unit.

[0016] In the above method implementations, the first measured stream can be the same as the first target stream, and the second measured stream can be the same as the second target stream. For example, the at least two physical variates can include temperature, vapour pressure, and flow rate, the predicted chemical variate can be solvent content, the first target stream can be the diluted bitumen overflow stream fed to the Solvent Recovery Unit (SRU), and the second target stream can be the bitumen product stream from the SRU, thereby assessing the efficiency of the SRU.

[0017] In combination with at least one of the above described implementations, the method can further include comparing the real-time value of the predicted chemical variate to a threshold value for the predicted chemical variate. Optionally, the method can further include producing a warning indicator when the real-time value of the predicted chemical variate deviates from the threshold value.

[0018] In the implementation where the real-time value of the predicted chemical variate deviates from the threshold value, the method can include executing a corrective action. The corrective action can be automated. The corrective action can further be a temporary corrective action and the method can include operating the temporary corrective action until a subsequent corrective action is triggered.

[0019] For example, the measured stream can be the same as the target stream, the target stream can be the bitumen product stream, the predicted chemical variate can be a solvent content of the target stream, and the corrective action can include recycling the bitumen product stream to the Solvent Recovery Unit (SRU).
Date Recue/Date Received 2020-10-09

[0020] In another example, the measured stream can be the same as the target stream, the target stream can be the bitumen product stream, the predicted chemical variate can be a solvent content of the first target stream, and the corrective action can include reducing a solvent to bitumen ratio of the solvent diluted bitumen froth fed to the Froth Separation Unit (FSU).

[0021] In another example, the measured stream can be the same as the target stream, the target stream can be the bitumen product stream from the Solvent Recovery Unit (SRU), the predicted chemical variate can be a solvent content of the target stream, and the corrective action can include recycling at least a portion of the bitumen product stream to the SRU.

[0022] In another example, the measured stream can be the same as the target stream, the target stream can be the diluted bitumen overflow stream from the Froth Separation Unit (FSU), the predicted chemical variate can be a residual metal content of the target stream, and the corrective action can include:
increasing or decreasing the dosage of an alkaline agent added to the bitumen-containing aqueous slurry;
substituting the alkaline agent for another alkaline agent of a different type;
adjusting a dosage of an inhibitor added to the bitumen-containing aqueous slurry; and modifying water dilution of the bitumen-containing aqueous slurry.

[0023] In the above implementation, the residual metal can include at least one of iron, calcium, sodium and magnesium. The threshold value of calcium can be below 10 ppm. The threshold value of sodium can be below 100 ppm.

[0024] In another example, the predicted chemical variate can be nickel and vanadium content, the measured stream can be the same as the target stream, the target stream can be the diluted bitumen froth stream that is fed to the Froth Separation Date Recue/Date Received 2020-10-09 Unit (FSU), and the corrective action can include adjusting the amount of a diluent added to the bitumen froth stream.

[0025] In another example, the predicted chemical variate can be nickel and vanadium content, the measured stream can be the same as the target stream, the target stream can be the diluted bitumen overflow stream from the Froth Separation Unit (FSU), and the corrective action can include adjusting asphaltene rejection in the FSU.

[0026] In another example, the predicted chemical variate can be nickel and vanadium content, the measured stream can be the same as the target stream, the target stream can be the diluted bitumen overflow stream from the Froth Separation Unit (FSU) or the bitumen product stream from the Solvent Recovery Unit (SRU), and the corrective action can include:
increasing the amount of solvent in the diluted bitumen froth stream fed to the FSU, thereby increasing a solvent-to-bitumen ratio in the FSU;
increasing asphaltene rejection in the FSU; or a combination thereof.

[0027] In the above implementation, the threshold value of nickel can be between 50 and 60 ppm. The threshold value of vanadium can be between 130 and 160 ppm.

[0028] In another aspect, there is provided a method for continuous soft-sensing of compositional characteristics of a PFT process stream in a secondary bitumen extraction operation. The method includes:
determining a real-time value of a predicted chemical variate of a target PFT
process stream by correlating at least two physical variates of a measured PFT

process stream, wherein the at least two physical variates of the measured PFT process stream are detected in real-time.

[0029] In some implementations of the method, the at least two physical variates can be a combination of pH, temperature, vapour pressure, density, viscosity, or flow rate.
The predicted chemical variate can be bitumen content, water content, solvent content, Date Recue/Date Received 2020-10-09 solids content, asphaltenes content, residual metals content, naphthenic acids content, total acid number (TAN), clay content, calcium content or fines content.

[0030] In combination with at least one of the above implementations, the measured PFT process stream can be the same as the target PFT process stream.

[0031] In combination with at least one of the above implementations, the predicted chemical variate and the at least two physical variates can be correlated using a multivariate model comprising multiple equations correlating multiple chemical variates to multiple physical variates, each physical variate being attributed a weighting in each equation of the multivariate model.

[0032] In some implementations, the method can further include determining a further dependent chemical variate based on the predicted chemical variate.

[0033] In combination with at least one of the above implementations, the target PFT
process stream can be a first target PFT process stream, the measured PFT
process stream can be a first measured PFT process stream, and the method can further include:
determining the real-time value of the predicted chemical variate of a second target PFT process stream by correlating the at least two physical variates of a second measured PFT process stream; and comparing the real-time values of the predicted chemical variate of the first target PFT process stream and of the second target PFT process stream, to assess the efficiency of a bitumen froth treatment unit.

[0034] In combination with at least one of the above implementations, the method can include comparing the real-time value of the predicted chemical variate to a threshold value for the predicted chemical variate. Optionally, the method can include producing a warning indicator when the real-time value of the predicted chemical variate deviates from the threshold value. Further optionally, the method can include executing a corrective action when the real-time value of the predicted chemical variate deviates from the threshold value. The corrective action can be automated. The corrective action can further be a temporary corrective action and the method can include operating the temporary corrective action until a subsequent corrective action is triggered.

Date Recue/Date Received 2020-10-09

[0035] In combination with at least one of the above implementations of the method, each of the target PFT process stream and the measured PFT process stream can be one of a bitumen froth stream from a Primary Separation Vessel (PSV), a diluted bitumen froth stream being fed to a Froth Separation Unit (FSU), a diluted bitumen overflow stream from the FSU, a bitumen product stream from a Solvent Recovery Unit (SRU), a recovered solvent stream from the SRU, and a tailings stream from a Tailings Solvent Recovery Unit (TSRU).[0035a] In one aspect, there is provided a method for producing bitumen comprising:
(i) producing multiple process streams from a bitumen-containing aqueous slurry, the multiple process streams comprising a target stream and a measured stream;
(ii) measuring real-time values of at least two physical variates of the measured stream, wherein the at least two physical variates are selected from the group consisting of pH, temperature, vapour pressure, density, viscosity and flow rate;
(iii) selecting a predicted chemical variate of the target stream based on the at least two physical variates that are measured, wherein:
the predicted chemical variate is asphaltenes content and the at least two physical variates comprise temperature, density and flow rate;
the predicted chemical variate is water content and the at least two physical variates comprise temperature, density and flow rate;
the predicted chemical variate is solids content and the at least two physical variates comprise temperature, density and flow rate;
the predicted chemical variate is asphaltenes content, water content, solids content or residual metals content and the at least two physical variates comprise temperature, flow rate and viscosity;
the predicted chemical variate is the total acid number (TAN), naphthenic acid content or calcium content, and the at least two physical variates comprise pH and temperature; or Date Recue/Date Received 2022-09-29 the predicted chemical variate is solvent content and the at least two physical variates comprise temperature, vapour pressure, and flow rate;
and (iv) determining a real-time value of the predicted chemical variate of the target stream by correlating the at least two physical variates.
[0035b] In another aspect, there is provided a method for continuous soft-sensing of compositional characteristics of a PFT process stream in a secondary bitumen extraction operation, the method comprising:
selecting a predicted chemical variate of a target PFT process stream based on the at least two physical variates of a measured PFT process stream, wherein:
the predicted chemical variate is asphaltenes content and the at least two physical variates comprise temperature, density and flow rate;
the predicted chemical variate is water content and the at least two physical variates comprise temperature, density and flow rate;
the predicted chemical variate is solids content and the at least two physical variates comprise temperature, density and flow rate;
the predicted chemical variate is asphaltenes content, water content, solids content or residual metals content and the at least two physical variates comprise temperature, flow rate and viscosity;
the predicted chemical variate is the total acid number (TAN), naphthenic acid content or calcium content, and the at least two physical variates comprise pH and temperature; or the predicted chemical variate is solvent content and the at least two physical variates comprise temperature, vapour pressure, and flow rate;
and determining a real-time value of the predicted chemical variate of the target PFT
process stream by correlating the at least two physical variates of the measured PFT process stream;
9a Date Recue/Date Received 2022-09-29 wherein the at least two physical variates of the measured PFT process stream are detected in real-time.
BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 is a process flow diagram showing primary and secondary extraction operations of bitumen from mined oil sands ore.

[0037] Figure 2 is a schematic illustrating a PFT operation including direct measurement of physical variates of PFT process streams via dedicated analyzers and meters, and prediction of chemical variates via soft sensors.

[0038] Figure 3 is a schematic illustrating an example multivariate model, including three soft sensors, for sensing three chemical variates based on a set of physical variates only.

[0039] Figure 4 is a graph representing correlated solvent content in a bitumen product stream according to three measured physical variates, namely temperature, flow rate and pressure.

[0040] Figure 5 is a graph representing correlated calcium content in a bitumen product stream according to measured pH in an aqueous slurry fed to a primary separation vessel (PSV).
DETAILED DESCRIPTION

[0041] Techniques described herein relate to methods of predicting certain variates of process streams generated during operations related to bitumen extraction. The predicted variates are chemical variates of the process streams, and the prediction is based on a multi-variate model having at least two physical variates as an input. The at 9b Date Recue/Date Received 2022-09-29 least one physical variate is directly measured from the process streams, via process instrumentation.

[0042] Figure 1 illustrates a general example process for primary and secondary extraction of bitumen from mined oil sands ore, as detailed below. It is noted that this is simply an example of a bitumen extraction operation and that variations of this example process can be used in conjunction with the soft sensor techniques described herein.
Primary bitumen extraction

[0043] Referring to Figure 1, oil sands ore 10 is crushed in a crushing unit 12 to obtain a crushed ore 13. The crushed ore 13 is then mixed with process water 14 (e.g., warm or hot water) in a mixing unit 16 to remove oversized clumps and form an aerated aqueous oil sands slurry 18. The mixing unit 16 can be, for instance, a rotary breaker that breaks up lumps of oil sands into smaller sized particles. The process water 14 and the sized oil sands material form the aqueous oil sands slurry 18, which can generally include between 5 wt% and 15 wt% bitumen, about 80 wt% solids, and between about 5 wt% and 15 wt% water, although other compositions are possible. The aqueous slurry 18 can then be shear conditioned to prepare the slurry for extraction of the bitumen from the solid minerals and water. The conditioning of the aqueous slurry 18 is typically performed through hydrotransport via a pipeline, which facilitates increased mixing, aeration and breakdown of lumps of oil sands ore in preparation for bitumen separation.

[0044] Still referring to Figure 1, the aqueous slurry 18, which can optionally be further diluted with process water 14, is transported to a primary separation vessel (PSV) 20. The PSV can also be referred to as a primary separation cell, "sep cell", or gravity separation cell. The PSV typically uses flotation and gravity mechanisms to separate bitumen from coarse sand and other solid particles, although other mechnicams and separator designs are possible. In the primary separation process, bitumen in the aqueous slurry 18 detaches from sand particles and attaches to air bubbles that are injected into the PSV 20, thereby facilitating bitumen droplets to rise and float to the top of the PSV 20, forming the primary bitumen froth 22 that is recovered typically as an overflow stream. Coarse particles contained in the aqueous slurry 18 are relatively heavy and tend to sink to the bottom of the PSV 20. The portion of the aqueous slurry 18 that is not heavy enough to sink to the bottom of the PSV 20 but not light enough to float Date Recue/Date Received 2020-10-09 tends to remain in the middle of the PSV 20, and can be referred to as middlings 26. The aqueous slurry 18 is thus separated into three streams withdrawn from the PSV:
a primary tailings underflow stream 24 (also referred to as coarse tailings), a middlings stream 26, and a bitumen froth overflow stream 22. The primary tailings 24 can then be disposed of in a tailings pond 50 or further treated to extract bitumen.

[0045] Still referring to the implementation illustrated in Figure 1, the middlings stream 26 can be sent to a secondary separation vessel 28, to be separated into a secondary bitumen froth 30 and secondary tailings 32. The secondary tailings 32 can also be referred to as a fine tailings stream herein as they contain higher fines content compared to the coarse tailings 24. The secondary bitumen froth 30 can be fed back to the primary separation vessel 20. Alternatively, the secondary bitumen froth 30 can be added directly to the primary bitumen froth 22. It is also noted that there may be additional separation vessels downstream of the secondary separation vessel 28, further enabling separation of residual bitumen from the water and mineral solids. The secondary tailings 32 can be disposed of in a tailings pond 50 or further treated to extract bitumen.

[0046] The bitumen froth overflow stream 22 recovered from the primary separation vessel is referred to as bitumen froth and typically includes about 60 wt%
bitumen, about 30 wt% water, and about 10 wt% solid materials, although these percentages can vary depending on various factors. The solid materials in the bitumen froth typically include hydrophilic mineral materials and heavy minerals, which can include adsorbed insoluble organic material. The primary tailings 24 and secondary tailings 32 generally include between about 45 wt% and about 55 wt% solid materials, between about 45 wt%
and about 55 wt% water, and residual bitumen (typically between about 1 wt% and about 3 wt% bitumen). The solid materials in the primary and secondary tailings 24, 32 are mainly sand and other fine hydrophilic mineral materials.

[0047] In some implementations, one or more alkaline agents, such as caustic soda (NaOH), sodium silicate, sodium bicarbonate, sodium phosphate, and the like, can be added directly to the aqueous slurry 18, before entering the PSV 20, to chemically condition and prepare the aqueous slurry 18 for bitumen extraction and separation in the PSV 20. The alkaline agent can be added to the process water 14, to the mixing unit 16, Date Recue/Date Received 2020-10-09 to the aqueous slurry before, during or after hydrotransport, and/or can be added directly into the PSV 20.
Secondary bitumen extraction

[0048] Referring to Figure 1, the bitumen froth 22 is further treated in a bitumen froth treatment operation 11 that includes several units. Bitumen froth 22 is first sent to a froth treatment process 34, also referred to as a froth separation unit (FSU), in which the bitumen froth 22 is diluted with solvent 36 to obtain a solvent diluted bitumen froth.

[0049] The solvent 36 can be a paraffinic solvent, which can for example include C4 to C8 aliphatic compounds and/or certain natural gas condensates. When a paraffinic solvent is used in conditions to induce some asphaltene precipitation, the bitumen froth treatment can be referred to as paraffinic froth treatment (PFT) and the process streams as PFT process streams. Pentane is one specific example of a paraffinic solvent that can be used in PFT operations. The paraffinic solvent is used under conditions such that, when added to the bitumen froth, the solvent induces precipitation of asphaltene-containing aggregates that trap water and fine mineral solids, thereby producing a "cleaner" bitumen component. Higher solvent-to-bitumen ratios tend to lead to higher levels of asphaltene precipitation from the bitumen.

[0050] Still referring to Figure 1, the diluted bitumen froth is separated in the FSU 34 into a diluted bitumen overflow 38 and froth treatment tailings 40 that include solid materials (hydrophilic mineral materials, heavy minerals and insoluble organic materials), water, residual diluent and residual bitumen. The diluted bitumen overflow 38 can be sent to a solvent recovery unit (SRU) 54, which produces two streams as recovered solvent 56 and a bitumen product 58. The FSU 34 itself can include two or three settler vessels that are arranged in a counter-current configuration (not illustrated in Figure 1), or other types of separation units.

[0051] In some implementations, the froth treatment tailings 40 are treated in an oil sands tailings treatment process 42, which can employ a tailings solvent recovery unit (TSRU), in order to separate the froth treatment tailings 40 into various recovered materials 44 such as solvent and an aqueous stream 46, also referred to as TRSU
tailings, including process water, heavy minerals, and/or hydrophilic mineral materials as Date Recue/Date Received 2022-09-29 well as residual bitumen and solvent. The TSRU tailings 46 including process water and hydrophilic mineral materials can be disposed of in a tailings pond 50 for settling.

[0052] In the implementation shown in Figure 1, the coarse tailings stream 24 and the fine tailings stream 32 are added to the TRSU tailings 46 for disposal in the tailings pond 50.

[0053] Still referring to Figure 1, an overlying water phase can be pumped out of the tailings pond 50 and re-used as recycled process water 52 in the mixing unit 16 to obtain the aqueous slurry 18, as well as in various other applications within the oil sands processing facility.

[0054] It should be noted that a "process stream" as used herein means any fluid stream involved in the bitumen extraction operation. The process streams can therefore include a bitumen-containing aqueous slurry, a bitumen froth stream, a diluted bitumen froth stream, a diluted bitumen overflow stream from the FSU, first or second stage overflow streams in a multi-stage FSU, first or second stage underflow streams in a multi-stage FSU, a recovered solvent stream from the SRU and TSRU, a bitumen product stream from the SRU, a TSRU tailings stream, as well as fluid streams such as diesel-containing streams used for start-up or cleaning the PFT vessels or lines. The process stream can be characterized as a multi-phase fluid containing a hydrocarbon phase, an aqueous phase and solids; a two-phase fluid; or a single-phase fluid in some cases. When the solvent used to produce the diluted bitumen froth stream is a paraffinic solvent, the process stream can be referred to as a "PFT process stream".

[0055] It should be noted that the solvent/diluent that can be used is not limited to a paraffinic solvent as described herein, and that particular implementations described herein in relation to PFT process streams or a paraffinic solvent can be applied to any process stream deriving from bitumen extraction or any solvent that is generally used in bitumen extraction, such as a naphtenic solvent.
Process stream variates

[0056] The method described herein includes real-time collection of physical variates values of a measured process stream. These collected values of physical variates can Date Recue/Date Received 2020-10-09 serve for further prediction of real-time values of at least one chemical variate of a target process stream, thereby referred to as the predicted chemical variate.

[0057] The "measured process stream" or "measured stream" refers herein to the process stream that is subjected to detection of at least two physical variates thereof via in-line process instrumentation. The terms "measured" and "detected" can be used herein interchangeably, as detection of the physical variates is performed via direct measurement of the physical variates with in-line analyzers and meters available to one skilled in the art.

[0058] The "target process stream" or "target stream" refers herein to the process stream that is subjected to the prediction of at least one chemical variate thereof via the soft sensor. In some implementations of the soft-sensing techniques described herein, the target stream can be the same as the measured stream. For example, both measured stream and target stream can be the diluted bitumen froth stream or the bitumen product stream. In other implementations of the soft-sensing techniques described herein, the target stream can be different from the measured stream.
For example, the measured stream can be the bitumen-containing aqueous slurry and the target stream can be the bitumen product stream.

[0059] In the PFT implementation illustrated in Figure 2, process instrumentation for measuring real-time values of the physical variates can include online analyzers and meters 60 that are positioned at various locations of a PFT operation. The analyzers and meters 60 can include multiple probes, each probe being dedicated to a specific measured PFT process stream (e.g. 22, 38, 46 or 58) and each being configured to measure at least one physical variate. The at least two physical variates that are used according to the present sof-sensing techniques data are directly measured in real-time via the online analyzers and meters 60.

[0060] Physical variates that are relevant to the bitumen froth treatment operation can include pH, temperature, vapour pressure, flow rate, density, and viscosity. For example, the method can include collecting real-time values of at least two physical variates of at least one of a measured stream being the bitumen-containing aqueous slurry, the bitumen froth, the solvent diluted bitumen froth, the diluted bitumen overflow, the bitumen product, the recovered solvent and/or the TSRU tailings.

Date Recue/Date Received 2020-10-09

[0061] It should be noted that residence time¨within lines or in particular vessels/units¨of a measured process stream can also be considered herein as a physical variate that can be used as input for a soft sensor.

[0062] In some implementations, the online analyzers and meters that are used for direct measurement of the physical variates include a coriolis flow meter, a point source nuclear density gauge (for slurry), a wedge flow meter (for slurry), a sonar flow meter (for non-intrusive measurement), an ultrasonic thermocouple in a thermowell, a handheld IR
meter (for spot temperature measurements), a pressure gauge (that measures the pressure of liquid and/or vapour), a differential pressure meter (that measures pressure drops across items such as packing modules or between vessels such as the FSU
and a froth surge vessel), or any combinations thereof. It is noted that Figure 2 is only provided as an example and it should be understood that the direct measurement of physical variates of measured process streams according to the present techniques can be performed by various measurement instrumentation and equipment.

[0063] The method further includes real-time prediction of at least one predicted chemical variate based on the at least two physical variates of the measured stream.
Still referring to Figure 2, the online analyzers and meters 60 can be in communication with a control unit 62 where the detected physical variates are treated as input for determination of the predicted chemical variate of the target PFT process stream. For example, the real-time values of the detected physical variates can be compiled in a routing information base (RIB) which stores data in a router or a network host. The control unit 62 can further include a central computing system that is programmed to to predict a chemical variate based on a combination of physical variates via a multivariate model.

[0064] Chemical variates as described herein can refer to compositional characteristics of the PFT process streams. Chemical variates can include bitumen content, solvent content, water content, solids content, asphaltenes content, clay content, calcium content, naphtenic acids content, total acid number (TAN), and residual metals content.

[0065] It should be noted that a "residual metal" as discussed herein refers to a metal, an alkaline earth metal or a metalloid that can be present in the bitumen froth Date Recue/Date Received 2020-10-09 treatment streams (including PFT process streams). Metals can be found in process water and in the oil sands ore used to prepare the oil sands slurry, such that residual metals are present in the process streams. More particularly, residual metals can include native materials present in the ore, including metals such as aluminum (Al), iron (Fe), nickel (Ni) and vanadium (V); metalloids such as silicon (Si); and alkaline earth metals such as calcium (Ca) and magnesium (Mg).

[0066] Soft-sensing one or more of the above-listed chemical variates can provide different indications and insights regarding bitumen product quality and upstream processing performance.

[0067] Information regarding the chemical composition of the various PFT
process streams can indeed be advantageous. For example, as precipitated asphaltenes can form a portion of the solvent froth treatment tailings stream 40, and in order to produce cleaner bitumen and at the same time to reduce asphaltenes rejection, controlling solvent-to-bitumen ratio in the FSU 34 can be critical for optimizing production and costs.
In addition, before the bitumen product is sent out to a pipeline, its quality and solvent content have to meet pipeline specifications. These specifications cover a variety of chemical properties, including residual solvent (e.g., < 800 ppm desirable), water (e.g., <
1000 ppm ), solids (e.g., <500 ppm), and asphaltenes (e.g., ¨12%), for example. Some specifications relate to product quality (e.g., water, solids, asphaltenes), whereas others relate to product safety (e.g., solvent concentration). Some specifications relate to sales market, e.g., metals content versus potential catalyst poisoning affecting premium status in refineries.

[0068] For example, the contents of residual metals in PFT process streams can vary from the start-up mode to the mature mode, or following the addition of a certain process-aid at a particular part of the process or into a particular froth treatment stream.
The calcium concentration can be up to 200 ppm in the bitumen product when caustic soda is added in primary extraction, whereas the calcium concentration can be below 10 ppm in the bitumen product when the slurry is untreated by caustic soda. In addition, some residual metals contents can remain substantially within the same range in start-up and mature modes. Nickel concentration in the bitumen product can be between 50 ppm and 60 ppm in both start-up mode and mature mode when the asphaltene content Date Recue/Date Received 2020-10-09 in the bitumen product is about 10 wt%. Thus, depending on the type of metal that is predicted, the stage or maturity of the extraction operation and other compositional features of the bitumen product, changes in metals concentration can provide valuable information for an operator.

[0069] In addition, dosing of the added alkaline agent can impact the content of certain residual metals which are present in certain downstream materials, such as final bitumen product. Metals content can influence bitumen product quality and can also be an indicator of upstream processing characteristics. For instance, when adding caustic soda for the PSV, ions exchange between Na + and Ca2+, leading to the formation of calcium naphthenates with the naturally occurring naphthenic acids in the oil sands.
Calcium naphthenates can be undesirable and can lower bitumen quality.
Elevated caustic soda levels can also lead to additional Ca2+ in the bitumen froth and can lead to the emulsification of bitumen and smaller bitumen droplets, which can impair bitumen recovery.
Soft sensing implementations

[0070] In contrast to techniques relying on manual sampling or specialized online chemical composition sensors (e.g., Near Infrared (NIR) analyzers) for measuring chemistry of the streams, the soft sensing techniques proposed herein rely on inferential measurement (i.e., via correlation equations of multivariate models) of chemical variates of target process streams deriving from bitumen extraction operations based on detected physical variates of measured process streams. The process streams (including target streams and measured streams) can be derived from primary extraction or secondary extraction. For example, the process streams can be PFT process streams. More specifically, at least one chemical variate of the target stream can be predicted based on at least two physical variates of the measured stream via a multivariate model.

[0071] Soft sensing can be defined as an indirect measurement, or a prediction, of a dependent variate based on directly measured variates. The soft sensor can derive from a multivariate model that is established for "sensing", via prediction, at least one sensed dependent variate. A soft sensed, sensed or predicted chemical variate can refer to an indirectly measured chemical variate, via a prediction based on the multivariate model using one or more directly measured physical variates. In the present context, Date Recue/Date Received 2020-10-09 sensed/predicted variates can be chemical variates of PFT process streams, which are correlated to one or more directly measured physical variates of such streams.

[0072] Referring to Figure 2, the control unit 62 can operate multiple soft sensors, each soft sensor being based on a multivariate model tailored to predict at least one chemical variate of the target stream that depends on a specific combination of physical variates of the measured stream, the physical variates being detected in real-time via the online analyzers and meters 60. The real-time values of the detected physical variates are treated as input of equations of the multivariate models that make up the soft sensors to estimate corresponding real-time values of predicted chemical variates of the PFT streams. The real-time values of the at least one predicted chemical variate can be displayed to an operator in various formats (e.g., graphs, tables, etc.) via the control unit 62.

[0073] Referring to Figure 3, in one implementation, the soft sensors can be defined by equations of the multivariate model having different input combinations of measured physical variates, depending on what is predicted as the chemical variate of the PFT
process stream. Each physical variate of the input combination is attributed an appropriate weighting in an equation of the multivariate model according to the dependent chemical variate to be predicted and the related target stream. For example, as seen in Figure 3, the input combination of collected physical variates can include temperature (T), vapour pressure (P), flow rate (F) and density (D), with each physical variate being attributed a specific weight (a, b, c, and d) in a given equation of the multivariate model. It should be noted that the attributed weight of a physical variate in the equation can be nul, as some measured physical variates of the monitored set may not be covariants of a given predicted chemical variate. For example, measured density (D) of the bitumen product stream can be given little or no weight (d1 = 0) in the first soft sensor equation that is used for sensing solvent content (chemical variate 1) in the same bitumen product stream, as solvent content is a chemical variate mainly dependent on temperature (T), vapour pressure (P), and flow rate (F) of the bitumen product stream.
Depending on the number of measured and collected physical variates via the existing process instrumentation, one or more physical variates of the input combination can be attributed a nul weight.

Date Recue/Date Received 2020-10-09

[0074] In some implementations, the input combination can include at least two physical variates, such as density and temperature, or pH and temperature. In further implementations, the input combination can include at least three physical variates, such as viscosity, temperature and flow rate.

[0075] For example, solvent content can be soft-sensed based on an input combination of measured temperature, vapour pressure, and flow rate. Referring to Figure 4, one can see that solvent content in the bitumen product stream is correlated to the temperature of that same stream, at substantially constant vapour pressure and flow rate.

[0076] In another example, asphaltene content can be soft-sensed based on an input combination of measured density, temperature, and flow rate. In another example, water content can be soft-sensed based on an input combination of measured density, temperature and flow rate. In another example, solids content can be soft-sensed based on an input combination of measured density, temperature and flow rate. In another example, naphthenic acids content and TAN can be soft-sensed based on an input combination of measured pH and temperature.

[0077] For example, the target stream can be at least one of a bitumen-containing aqueous slurry stream, a diluted bitumen froth stream, a diluted bitumen overflow stream, a bitumen product stream, a recovered solvent stream and a TSRU
tailings stream. For example, the measured stream can be at least one of an aqueous slurry stream, a diluted bitumen froth stream, a diluted bitumen overflow stream, a bitumen product stream, a recovered solvent stream and a TSRU tailings stream. It should be noted that the target stream can be the same as the measured stream. In other words, the method can detect physical variates and predict at least one chemical variate from a same process stream. It should further be noted that the target stream can be different from the measured stream. For example, the physical variates of a measured PFT

process stream from the primary extraction stage can be used as input of a soft sensor to predict at least one chemical variate of a target PFT process stream of secondary extraction stage. More specifically, pH of the aqueous slurry entering PSV, at substantially constant temperature, can for example be used to predict TAN, naphthenic Date Recue/Date Received 2020-10-09 acid content and/or calcium content of the bitumen product stream generated by secondary extraction.

[0078] In addition, physical variates can be selected depending on the measured process stream. For example, viscosity is more easily measured for the bitumen product stream whereas density could be selected for measurements of the diluted bitumen overflow stream. In some implementations, the measured process stream can be the same as the target process stream, the target process stream can be a bitumen product stream, the predicted chemical variate is asphaltenes, water, solids or residual metals content and the set of physical variates comprises temperature, flow rate and viscosity of the bitumen product stream. It is also noted that the weightings and equations used to predict a given chemical variate could be different depending on the measured process stream from which the at least one physical variate is obtained.

[0079] In some implementations, the method can further include determining another dependent chemical variate based on the soft sensed chemical variate. For example, the method to produce can be programmed to produce residual metals content (e.g., nickel and vanadium content) based on the previously soft sensed asphaltene content of a PFT
process stream.

[0080] In order to evaluate the bitumen product quality and process efficiencies, chemical variates of target PFT process streams can be predicted in real-time based on the soft sensing techniques described herein. Referring to Figure 2, PFT
process streams of interest can include the diluted bitumen overflow stream 38 (immediately after the FSU 34 (38a) and immediately before SRU 54 (38b)), and the hot bitumen product 58 after the SRU 54. Referring to Table 1 below, chemical variates of interest are predicted via soft sensors for these PFT streams, using the above-identified input combinations of measured physical variates with appropriate weighting under a tailored multivariate model.
Table 1. Example of chemical variates of three PFT streams predicted by soft sensors Predicted PFT Unit FSU SRU
by Soft PFT stream Overflow Feed Product Sensor 38a 38h Date Recue/Date Received 2022-09-29 Bitumen Solvent Water Chemical Solids variates Asphaltenes Naphthenic Acids, TAN
Ni, V

[0081] In some implementations, the method can include soft-sensing the solvent content in the bitumen product stream, soft-sensing asphaltenes content in the diluted bitumen overflow stream, soft-sensing water content in the diluted bitumen overflow stream, and/or soft-sensing solids content in the diluted bitumen overflow stream.

[0082] It should be noted that predicted chemical variates and bitumen froth treatment streams other than those noted above, can be the subject the soft-sensing techniques. For example, the method can include soft-sensing of asphaltenes, solvent and clay contents in TSRU tailings.

[0083] It should further be noted that the soft-sensing techniques described herein can be applied to process streams of PFT operations, to Naphtenic Froth Treatment (NFT) operations, and to ore composition analysis. For example, soft sensing techniques as described herein can be applied to sensing, via prediction, clay content, fines content and connate water content of mined oil sands ore.

[0084] Soft sensing according to the techniques described herein facilitates continuous and reliable monitoring of chemical variates of the process streams, thereby facilitating real-time detection of off-specification compositions of such streams, for bitumen quality control and planning of process control strategies. The use of soft sensors further improves safety in the bitumen froth treatment facility, e.g.
by reducing the number of times a pressure envelope is broken for taking samples and by reducing the amount of off-specification streams at the facility.
Building the multivariate models

[0085] In some implementations, each equation of the multivariate models defining the soft sensors can be developed based on a multivariate analysis of historical data of Date Recue/Date Received 2020-10-09 each chemical variate for various target process streams. The historical data can be acquired by various methods, e.g., NIR-acquired data.

[0086] More particularly, the multivariate model equations of the soft sensor can be developed by collecting historic data on the chemical composition of the process streams using NIR (or other) online analyzers, and correlating with values of the physical variates that were collected by process instrumentation at the same time to develop multivariate model equations for chemical composition prediction. NIR-acquired data of chemical composition can thereby allow to develop the correlation equations with the values of the physical variates of given target streams.
Process control implementations

[0087] Process control strategies can be implemented based on the sensed chemical variates of the target process streams. More particularly, implemented process control strategies can include producing a warning indicator, subsequent to monitoring a deviation in at least one sensed chemical variate. The warning indicator can be a visual signal, an alarm, a notification or any combinations thereof. The warning indicator is produced to inform a selected receiving party (e.g., operation, maintenance, engineering or other selected parties, individually or as a group) that at least one chemical variate has been sensed to deviate from specification or to reach a threshold value.

[0088] Referring to Figure 2, the control unit 62 can display one or more warning notification(s) when at least one sensed chemical parameter is atypical. For example, process control can include sensing asphaltenes, water, and solids contents in the diluted bitumen overflow stream 38 to produce a warning notification signaling efficiency problems in the FSU 34, when the sensed contents approch or reach corresponding threshold values. In another example, the process control can include sensing solvent content in the diluted bitumen overflow stream 38 and in the bitumen product 58 for comparison thereof, to produce a warning notification when the efficiency of the SRU 54 is assessed as approaching or reaching a threshold value.

[0089] Process control strategies can be implemented based on sensed chemical variates of the PFT streams to predict the bitumen product quality. Final bitumen product in a PFT operation can require to meet quality specifications before being commercialized to refineries. Various residual elements considered as contaminants can Date Recue/Date Received 2020-10-09 be sensed by the present techniques to predict bitumen quality before being sent to refineries.

[0090] One or more alkaline agents, such as caustic soda (NaOH), sodium silicate, sodium bicarbonate, sodium phosphate and the like, can be added directly to the bitumen-containing aqueous slurry, before starting primary separation in the PSV, to chemically condition and prepare the aqueous slurry for bitumen extraction and separation in the PSV. Dosing of the added alkaline agent can impact the content of certain residual metals which are present in certain downstream materials, such as the bitumen product. Metals content can influence bitumen quality and can also be an indicator of upstream processing characteristics.

[0091] For example, the bitumen product could be required to contain less than 10 ppm of calcium. Referring to Figure 2, the control unit 62 can display the profile of calcium concentration that is sensed in diluted bitumen overflow 38, via a tailored soft sensor, to predict the bitumen product quality for example. In another example, the bitumen product could be required to comply with specifications with respect to vanadium and nickel contents. The content specification for nickel can be between 50 and 60 ppm, and the content specification for vanadium can be between 130 and ppm.

[0092] Residual metals can be found in the process water. Process water chemistry can evolve over time, from the moment a plant is put into operation and fresh water is used initially and in start-up processes, to many years later when processes have reached an equilibrium in terms of recycled process water that has gone through multiple cycles of separation processes. As seen in Figure 1, water from the tailings pond 50 can be reused as process water 52 to mix with the oil sands ore 13 and produce the oil sands slurry 18. This recycled process water can have a different water chemistry compared to fresh water. After a certain number of years of a plant's operation, i.e., once the plant could be said to be "mature", process water chemistry can reach an equilibrium stage. In contrast to a mature plant, water chemistry of process water used at a newer plant can change substantially in the first few years of operation, in particular with regard to the residual metals.

Date Recue/Date Received 2020-10-09

[0093] The contents of vanadium and nickel in the produced bitumen can also be indicative of the asphaltene rejection. Indeed, nickel and vanadium are heavier metals which tend to be agglomerated with asphaltenes. In addition, nickel and vanadium are known to be poisonous to the catalyst used in refineries. Therefore, prediction of the nickel and vanadium contents in the produced bitumen via a tailored soft sensor can be indicative of the bitumen quality, and validates whether the bitumen meets the specifications of refineries.

[0094] In some implementations, process control strategies can include producing the warning indicator subsequent to monitoring a deviation in at least one sensed chemical variate, and further performing a corrective action in response to the monitored deviation. For example, the corrective action can be an automated corrective action triggered by the control unit 62. In another example, the corrective action can be a temporary corrective action that is performed until another corrective action is triggered by an operator upon evaluation of the warning indicator. Referring to Figure 2, the control unit 62 can be further programmed to actuate the corrective action to a particular unit of the PFT operation, when sensed chemical variates have values that are reaching a given threshold value.

[0095] The multiple operating parameters that can be controlled in response to the difference between a sensed chemical variate and a chemical variate specification include but are not limited to a dosage of an alkaline agent added to the bitumen-containing aqueous slurry, a nature of the alkaline agent, a dosage of an inhibitor added to the bitumen-containing aqueous slurry, a water dilution of the bitumen-containing aqueous slurry, a temperature of the FSU, a nature of the solvent/diluent (e.g. paraffinic solvent) added to the bitumen froth stream, a bitumen content of the bitumen froth stream prior to the FSU, a settling parameter of the FSU, a solvent to bitumen ratio of the solvent diluted bitumen froth, or a combination thereof.

[0096] For example, process control strategies can include sensing solvent content in the bitumen product 58 based on a combination of measured physical variates, and the automated corrective action can be recycling the bitumen product 58 to the if the sensed solvent content is higher than a threshold value. The recycling of the bitumen product 58 to the SRU 54 can be performed temporarily, until another corrective Date Recue/Date Received 2020-10-09 action can be evaluated and triggered, such as a corrective modification in the solvent to bitumen ratio of the solvent diluted bitumen froth fed to the FSU 34. In another example, process control strategies can include sensing nickel and vanadium content in the bitumen product 58 based on a combination of measured physical variates, and the corrective action can be increasing solvent addition to provide a higher solvent-to-bitumen ratio in the solvent-diluted bitumen froth that is fed to the FSU 34 to increase asphaltene precipitation, if the sensed content in nickel and vanadium is above a given threshold.

[0097] The additional corrective action, in contrast to the automated preliminary corrective action, can be evaluated and triggered by an operator or engineer when notified by the warning indicator displayed on the the display unit 66 of the central computing system 64.

[0098] Other process control strategies can be implemented based on the sensed chemical variates of the primary extraction streams. The process control can include soft-sensing residual metals in the bitumen product to determine bitumen quality, and further adjust dosage of process-aids, such as alkaline agent or inhibitors, in primary extraction operations and froth treatment operations if the residual metal contents are deviating and even off-limit.

[0099] Process-aids, such as an alkaline agent (e.g. caustic soda), can be added to the oil sands slurry in primary extraction operation. When adding caustic soda to the slurry for the PSV, ions exchange between Na + and Ca2+, leads to the formation of calcium naphthenates with the naturally occurring naphthenic acids in the oil sands.
Calcium naphthenates can be undesirable and lower bitumen quality. Thus, according to the present techniques, calcium content of PFT process streams can be predicted via soft sensing, to further control the dosage of alkaline agents such as caustic soda in primary extraction. Referring to Figure 2, the control unit 62 can display the profile of predicted calcium concentration in diluted bitumen overflow 38. For example, calcium content of the diluted bitumen overflow stream (or of the bitumen product) could be predicted based on pH, temperature and residence time of the aqueous slurry between caustic addition and PSV. Figure 5 shows an example of the correlation between calcium in the bitumen product stream and the pH of the aqueous slurry in primary Date Recue/Date Received 2020-10-09 extraction. Referring to Figure 2, the control unit 62 can produce a warning indicator when the predicted real-time value reaches a given threshold. Different corrective actions can further be undertaken to adjust the calcium concentration. For example, the corrective actions can include changing caustic soda for an alternative alkaline agent that is added to the bitumen-containgin aqueous slurry, altering the composition of the alkaline agents or general process aids added to the slurry, adjusting the amount of added caustic soda, or controlling the flow rate of dilution water that is added to the slurry before being fed into the PSV, so as to prevent a caustic overdose which could reduce bitumen product quality for example.

[00100] It should be noted that inhibitors of the calcium naphthenates can be used for addition in the aqueous slurry. The process control can thus include predicting calcium content of the diluted bitumen overflow (or bitumen product) and further control alkaline agent addition and/or inhibitor addition into the aqueous slurry before PSV, when the calcium content deviates from specification or reach a threshold value.

[00101] It should be noted that various chemical elements of the PFT
streams, such as magnesium, iron, sodium, chloride and sulphate, can be sensed and relied on for process control. For example, sodium content of the bitumen product can be controlled under 100 ppm. When using of the presently described soft sensing techniques, compliance with given thresholds can be continuously controlled by predicting selected chemical variates of the PFT process streams. According to an advanced automated process, the control unit 62 can be configured as a multi-parameter control system which receives multiple input physical variates, determines multiple output chemical variates and controls multiple operating parameters of the primary extraction operation and of the PFT operation as exemplified in Figure 1. The multiple input physical variates are derived from direct online measurements taken from one or more PFT streams (or locations) in the facility; and the multiple output chemical variates are derived from real-time prediction via a soft sensor tailored to each chemical variate.

[00102] Once a deviation is monitored in at least one sensed chemical variate, the control unit 62 can trigger multiple temporary automated corrections to ensure, for example, that mass and energy balances are maintained for either individual components or bulk of the PFT process streams. The control unit 62 can temporarily Date Recue/Date Received 2020-10-09 proceed to multiple temporary automated corrections of operational parameters to ensure that efficiency of the process, e.g. production rate of the bitumen product, is maintained until appropriate corrective action(s) can be assessed and applied.

[00103] In some implementations, the control process can include assessment of the corrective actions either via algorithms or via a group of independent simplistic calculations. For example, algorithms can be used in the context of Artificial Intelligence (Al) to provide advanced diagnostics via machine learning that can be displayed to an operator or engineer. Such computer-aided diagnostic algorithms are able to assess corrective actions with respect to PFT streams. In addition, self-checking algorithms can be implemented to identify alternative control groups of physical variates that were not initially considered to sense given chemical variates. Based on the advanced diagnostic, the control unit 62 can implement automated corrective action with either no operational intervention or requiring acknowledgement from the operation personnel e.g.
management, operator, engineers etc. individually or as a group.

[00104] In other implementations, the control process can include triggering corrective action upon implementing modifications in the variate specifications and/or production requirements. Computer-aided diagnostic algorithms and appropriate error data can operate one or more corrective actions with respect to operational parameters of the PFT streams to ensure all production requirements and variate specifications (contractual, environmental, etc.) are within the newly imposed tolerances.
For example, the central computing system can maintain an asphaltenes content of the diluted bitumen overflow stream within new tolerances, while keeping bitumen production rate consistent, by modifying a solvent to bitumen ratio in the solvent diluted bitumen froth.
Such advanced control can ensure the asphaltenes to remain below a maximum value and not affect pumping requirements for the pipeline.

[00105] Advantageously, the use of the present advanced control process can help to ensure control decisions are made with a more complete dataset and allow better triaging of maintenance and operational loads. The PFT operation can therefore stably perform and safety can be enhanced, while prioritizing the workload for operational and maintenance staff.

Date Recue/Date Received 2020-10-09

[00106] In other implementations, additional online NIR-based measurement of the chemical variates can be performed to validate, via comparison, the values predicted via the soft sensors. The additional online NIR-acquired measurements can allow to continuously optimize the correlation equations of the multivariate model.

[00107] Obtaining NIR spectral measurement can include the use of an NIR
probe. In some implementations, at least one NIR probe is installed online, positioned in an upper region of a horizontal pipe section and within a hydrocarbon stratum; and a light source (e.g., laser beam) is emitted by the NIR probe into the PFT process stream.
The probe may be a reflectance probe or a transmission probe, and can be selected depending on the nature of the PFT process stream and the variate to be determined. The light emitted by the NIR probe interacts with the PFT process stream and the resulting radiation is captured by an NIR detector. The radiation received after interaction with the PFT process stream is captured and can be analysed by an NIR analyser, which provides the NIR spectral measurements. Any NIR analyser fitted with a fiber optic probe can be used to analyse the detected IR radiation and provide the NIR spectral measurements. For example, a Matrix-F FT-NIR spectrometer (Brukere) with transmission and reflectance probes may be used to take NIR spectral measurements.
In some implementations, the NIR spectral measurements are continuously obtained during operation of the bitumen froth treatment, and sent to the central computing system for validation of the predicted chemical variates. It should be noted that the physical variates could also be measured online via NIR analyzers.

Date Recue/Date Received 2020-10-09

Claims (44)

1. A method for producing bitumen comprising:
(i) producing multiple process streams from a bitumen-containing aqueous slurry, the multiple process streams comprising a target stream and a measured stream;
(ii) measuring real-time values of at least two physical variates of the measured stream, wherein the at least two physical variates are selected from the group consisting of pH, temperature, vapour pressure, density, viscosity and flow rate;
(iii) selecting a predicted chemical variate of the target stream based on the at least two physical variates that are measured, wherein:
the predicted chemical variate is asphaltenes content and the at least two physical variates comprise temperature, density and flow rate;
the predicted chemical variate is water content and the at least two physical variates comprise temperature, density and flow rate;
the predicted chemical variate is solids content and the at least two physical variates comprise temperature, density and flow rate;
the predicted chemical variate is asphaltenes content, water content, solids content or residual metals content and the at least two physical variates comprise temperature, flow rate and viscosity;
the predicted chemical variate is the total acid number (TAN), naphthenic acid content or calcium content, and the at least two physical variates comprise pH and temperature; or the predicted chemical variate is solvent content and the at least two physical variates comprise temperature, vapour pressure, and flow rate;
and (iv) determining a real-time value of the predicted chemical variate of the target stream by correlating the at least two physical variates.

2. The method of claim 1, wherein producing multiple process streams from the bitumen-containing aqueous slurry comprises separation of a bitumen froth stream and addition of a paraffinic solvent to the bitumen froth stream, and each of the multiple process streams is a paraffinic froth treatment (PFT) process stream.

3. The method of any one of claims 1 or 2, wherein the measured stream is the same as the target stream.

4. The method of claim 3, wherein the target stream is a diluted bitumen overflow stream from a Froth Separation Unit (FSU), the predicted chemical variate is the asphaltenes content of the diluted bitumen overflow stream, and the at least two physical variates comprise the temperature, the density and the flow rate.

5. The method of claim 3, wherein the target stream is a diluted bitumen overflow stream from a Froth Separation Unit (FSU), the predicted chemical variate is the water content of the diluted bitumen overflow stream, and the at least two physical variates comprise the temperature, the density and the flow rate.

6. The method of claim 3, wherein the target stream is a diluted bitumen overflow stream from a Froth Separation Unit (FSU), the predicted chemical variate is the solids content of the diluted bitumen overflow stream, and the at least two physical variates comprise the temperature, the density and the flow rate.

7. The method of claim 3, wherein the target stream is a bitumen product stream from a Solvent Recovery Unit (SRU), the predicted chemical variate is the asphaltenes content, the water content, the solids content or the residual metals content of the bitumen product stream, and the at least two physical variates comprise the temperature, the flow rate and the viscosity.

8. The method of claim 1 or 2, wherein the target stream is different from the measured stream.

9. The method of claim 8, wherein the target stream is a recovered solvent stream from a Solvent Recovery Unit (SRU), the measured stream is a bitumen product stream from the SRU, the predicted chemical variate is the solvent content of the solvent recovery stream, and the at least two physical variates comprise the temperature, the vapour pressure, and the flow rate.

10. The method of claim 8, wherein the target stream is a bitumen product stream from a Solvent Recovery Unit (SRU), the measured stream is the bitumen-containing aqueous slurry, the predicted chemical variate is the TAN, the naphthenic acid content or the calcium content, and the at least two physical variates comprise the pH and the temperature.

11. The method of any one of claims 1 to 10, wherein the predicted chemical variate and the at least two physical variates are correlated using a multivariate model comprising multiple equations correlating multiple chemical variates to multiple physical variates, each physical variate being attributed a weighting in each equation of the multivariate model.

12. The method of any one of claims 1 to 11, further comprising determining a further dependent chemical variate based on the predicted chemical variate.

13. The method of claim 12, wherein the predicted chemical variate is the asphaltenes content and the further dependent chemical variate is nickel and vanadium content.

14. The method of any one of claims 1 to 13, wherein the real-time values of the at least two physical variates are detected using at least one in-line sensor.

15. The method of claim 1 or 2, wherein the target stream is a first target stream, the measured stream is a first measured stream, the multiple process streams further comprise a second target stream and a second measured stream, and the method further comprises:
(iv) measuring the real-time value of each of the at least two physical variates of the second measured stream; and (v) determining the real-time value of the predicted chemical variate of the second target stream by correlating the at least two physical variates of the second measured stream; and (vi) comparing the real-time value of the predicted chemical variate of the first target stream determined via step (iii) with the real-time value of the predicted chemical variate of the second target stream determined via step (v), to assess the efficiency of a bitumen froth treatment unit.

16. The method of claim 15, wherein the first measured stream is the same as the first target stream, and the second measured stream is the same as the second target stream.

17. The method of claim 16, wherein the at least two physical variates comprise the temperature, the vapour pressure, and the flow rate, the predicted chemical variate is the solvent content, the first target stream is a diluted bitumen overflow stream fed to a Solvent Recovery Unit (SRU), and the second target stream is a bitumen product stream from the SRU, thereby assessing the efficiency of the SRU.

18. The method of claim 1 or 2, further comprising comparing the real-time value of the predicted chemical variate to a threshold value for the predicted chemical variate.

19. The method of claim 18, further comprising producing a warning indicator when the real-time value of the predicted chemical variate deviates from the threshold value.

20. The method of claim 18 or 19, further comprising executing a corrective action when the real-time value of the predicted chemical variate deviates from the threshold val ue.

21. The method of claim 20, wherein the corrective action is automated.

22. The method of claim 20 or 21, wherein the corrective action is a temporary corrective action and the method further comprises operating the temporary corrective action until a subsequent corrective action is triggered.

23. The method of any one of claims 20 to 22, wherein the measured stream is the same as the target stream, the target stream is a bitumen product stream from a Solvent Recovery Unit (SRU), the predicted chemical variate is the solvent content of the target stream, and the corrective action comprises recycling the bitumen product stream to the Solvent Recovery Unit (SRU).

24. The method of any one of claims 20 to 22, wherein the measured stream is the same as the target stream, the target stream is a bitumen product stream from a Solvent Recovery Unit (SRU), the predicted chemical variate is the solvent content of the first target stream, and the corrective action comprises reducing a solvent-to-bitumen ratio of a solvent diluted bitumen froth fed to a Froth Separation Unit (FSU).

25. The method of any one of claims 20 to 22, wherein the measured stream is the same as the target stream, the target stream is a bitumen product stream from a Solvent Recovery Unit (SRU), the predicted chemical variate is the solvent content of the target stream, and the corrective action comprises recycling at least a portion of the bitumen product stream to the SRU.

26. The method of any one of claims 20 to 22, wherein the measured stream is the same as the target stream, the target stream is a diluted bitumen overflow stream from a Froth Separation Unit (FSU), the predicted chemical variate is the residual metal content of the target stream, and the corrective action comprises:
increasing or decreasing the dosage of an alkaline agent added to the bitumen-containing aqueous slurry;
substituting the alkaline agent for another alkaline agent of a different type;

adjusting a dosage of an inhibitor added to the bitumen-containing aqueous slurry;
modifying water dilution of the bitumen-containing aqueous slurry; or any combinations thereof.

27. The method of claim 26, wherein the residual metal comprises at least one of iron, calcium, sodium and magnesium.

28. The method of claim 27, wherein the threshold value of calcium is below 10 ppm.

29. The method of claim 27 or 28, wherein the threshold value of sodium is below 100 ppm.

30. The method of any one of claims 20 to 22, wherein the predicted chemical variate is nickel and vanadium content, the measured stream is the same as the target stream, the target stream is a diluted bitumen froth stream that is fed to a Froth Separation Unit (FSU), and the corrective action comprises adjusting the amount of a diluent added to a bitumen froth stream.

31. The method of any one of claims 20 to 22, wherein the predicted chemical variate is nickel and vanadium content, the measured stream is the same as the target stream, the target stream is a diluted bitumen overflow stream from a Froth Separation Unit (FSU), and the corrective action comprises adjusting asphaltene rejection in the FSU.

32. The method of any one of claims 20 to 22, wherein the predicted chemical variate is nickel and vanadium content, the measured stream is the same as the target stream, the target stream is a diluted bitumen overflow stream from a Froth Separation Unit (FSU) or a bitumen product stream from a Solvent Recovery Unit (SRU), and the corrective action comprises:
increasing the amount of solvent in a diluted bitumen froth stream fed to the FSU, thereby increasing a solvent-to-bitumen ratio in the FSU;

increasing asphaltene rejection in the FSU; or a combination thereof.

33. The method of any one of claims 30 to 32, wherein the threshold value of nickel is between 50 and 60 ppm.

34. The method of any one of claims 30 to 33, wherein the threshold value of vanadium is between 130 and 160 ppm.

35. A method for continuous soft-sensing of compositional characteristics of a PFT
process stream in a secondary bitumen extraction operation, the method comprising:
selecting a predicted chemical variate of a target PFT process stream based on the at least two physical variates of a measured PFT process stream, wherein:
the predicted chemical variate is asphaltenes content and the at least two physical variates comprise temperature, density and flow rate;
the predicted chemical variate is water content and the at least two physical variates comprise temperature, density and flow rate;
the predicted chemical variate is solids content and the at least two physical variates comprise temperature, density and flow rate;
the predicted chemical variate is asphaltenes content, water content, solids content or residual metals content and the at least two physical variates comprise temperature, flow rate and viscosity;
the predicted chemical variate is the total acid number (TAN), naphthenic acid content or calcium content, and the at least two physical variates comprise pH and temperature; or the predicted chemical variate is solvent content and the at least two physical variates comprise temperature, vapour pressure, and flow rate;
and determining a real-time value of the predicted chemical variate of the target PFT
process stream by correlating the at least two physical variates of the measured PFT process stream;
wherein the at least two physical variates of the measured PFT process stream are detected in real-time.

36. The method of claim 35, wherein the measured PFT process stream is the same as the target PFT process stream.

37. The method of claim 35 or 36, wherein the predicted chemical variate and the at least two physical variates are correlated using a multivariate model comprising multiple equations correlating multiple chemical variates to multiple physical variates, each physical variate being attributed a weighting in each equation of the multivariate model.

38. The method of any one of claims 35 to 37, further comprising determining a further dependent chemical variate based on the predicted chemical variate.

39. The method of any one of claims 35 to 38, wherein the target PFT process stream is a first target PFT process stream, the measured PFT process stream is a first measured PFT process stream, and the method further comprises:
determining the real-time value of the predicted chemical variate of a second target PFT process stream by correlating the at least two physical variates of a second measured PFT process stream; and comparing the real-time values of the predicted chemical variate of the first target PFT process stream and of the second target PFT process stream, to assess the efficiency of a bitumen froth treatment unit.

40. The method of any one of claims 35 to 39, further comprising comparing the real-time value of the predicted chemical variate to a threshold value for the predicted chemical vari ate.

41. The method of claim 40, further comprising producing a warning indicator when the real-time value of the predicted chemical variate deviates from the threshold value.

42. The method of claim 40 or 41, further comprising executing a corrective action when the real-time value of the predicted chemical variate deviates from the threshold value.

43. The method of claim 42, wherein the corrective action is automated.

44. The method of claim 43, wherein the corrective action is a temporary corrective action and the method further comprises operating the temporary corrective action until a subsequent corrective action is triggered.