Blending

Blending back to whole crude

To calculate the properties of different streams in the refinery, AVEVA Unified Supply Chain recuts the feedstocks to the current model to match the cutting scheme in the refinery. So, to calculate the sulfur content of the products from crude distillation, AVEVA Unified Supply Chain uses the sulfur profile of each potential feedstock and calculates the sulfur content of the products that would be derived from this.

The total amount of many properties is conserved through the plant. For example, the mass of the materials cannot change, so the mass of feed must equal the mass of products. Similarly, sulfur cannot be created or destroyed in the refinery, so for a single feedstock the combined sulfur content of the products must be the same as the sulfur content of the feedstock, once the relative yield of each product is taken into account.

Properties such as this are said to blend back to whole crude. For a single feedstock, the weighted sum of the product property values must be equal to the original feedstock's property value. Common properties which blend back to whole crude include the following:

• Sulphur (Total)

• Nitrogen (Total)

• Nitrogen (Basic)

• Vanadium

• Nickel

Other properties do not blend back to whole crude. That is, the whole crude property value is not the weighted average of the property values for the feedstock products. For example, the whole crude pour point is not the weighted average of the pour points of the crude's products.

Common properties which do not blend back to whole crude include the following:

• Density

• Pour Point

• Viscosity

• Reid Vapour Pressure

  Note: Blending back of density to whole crude does not occur because of volume expansion.

  The table below shows the cut properties from a particular crude assay.

  Sample ID			2005FED
  Crude ID			Penara
  			350-370°C	370-380°C	380-450°C	450-540°C
  Mass Yield	D2892	%mass	3.19	1.73	12.92	10.62
  Sulphur (Total)	D4294	%mass	0.0596	0.0492	0.0444	0.0709
  Pour Point	D97	°C	33	39	48	54

Calculate the sulfur content of the 350-380°C cut

(3.19 * 0.0596 + 1.73*0.0492) / (3.19 + 1.73) = 0.0559 %

Calculate the pour point of the 370-450°C cut

Use the pour point blending index exp(0.03*Pour in °F)

exp(0.03*102.2)=21.46; exp(0.03*118.4)=34.33

(1.73 * 21.46 + 12.92 * 34.88) / (1.73 + 12.92) = 33.30

ln(33.30)/0.03 = 116.8°F = 47.1° C

Re-blending narrow cuts into wider streams

Some crude properties require a blending index to calculate the property value for a mix of different crudes. This rule also applies to fractions of the crude oil. For example, a refinery may produce an atmospheric gas oil fraction between 250-350°C, or it may produce a light and heavy gas oil fraction with cut points of 250-300°C and 300-350°C respectively. The yield of the atmospheric gas oil is obviously the sum of the yields of the light and heavy gas oil, and the sulfur content of the AGO is the weighted average of the sulfur contents of the LGO and HGO. However, to calculate the pour point of the AGO we need to use the pour point index of the LGO and HGO to determine the weighted value.

Unlike for whole crude properties though, it can be expected that when using the blending rule for a property the weighted average of a cut's constituent parts should equal the property value for the original cut. So for example, when using the pour point index blending rule, the weighted average of the LGO and HGO pour point must equal the AGO pour point.

This is true as long as the same pour point index is used to recut the original crude oil property profile to determine the cut property values as is used to re-blend the cut property values. To determine the property value for a cut, AVEVA Unified Supply Chain must recut the original property profile for that property. If it is a non-linear blending property it must use a blending rule in order to do this recutting. So, to calculate the pour point of the LGO, HGO and AGO, AVEVA Unified Supply Chain would use the pour point profile of the original feedstock, and a blending rule, to determine the pour point of the individual cuts. Once the pour point of the individual cuts is known these can be averaged using the pour point blending rule to calculate the index. To transform the calculated average index back to the original property value the inverse of the index would be used. In order to get the same original property value it is therefore necessary to use the same blend rule formula for each stage of the recutting and blending. If different blend rule formulae are used at different stages there is no guarantee that the properties will eventually balance because each blend rule is only an approximation to the true behavior.

Properties where consistency in blend rules is important for property balancing include the following:

• Pour Point

• Cloud Point

• Freeze Point

• D86 boiling point at X%

• Cetane Index

Properties such as Cetane Index will not blend back simply because they are calculated using a correlation which itself depends on another non-linear blending property. For example, cetane index is calculated from the D86 boiling point at 50%. Thus the cetane index of the blended LGO and HGO stream will not be the same as the cetane index of the AGO stream unless the same blending rule is used through all stages of the recutting and blending process.

Blend indices in process models

Base + Delta process models predict property values, and these predictions may be for non-linear properties. For example, a visbreaker may predict the viscosity of its products, visbreaker gasoil and visbreaker pitch. As viscosity is a non-linear blending property, it typically has a blend rule associated with it, and for later stream blending it is required that the stream viscosities are all in linear blend index space. AVEVA Unified Supply Chain tries to avoid unnecessary conversion between engineering (real) space and blend index space where possible. It does this using several methods:

• Where a property prediction is entered in engineering space but is a constant, it is automatically converted into blend index space for optimization. So a constant visbreaker gas oil viscosity measured in cSt would be converted into its blend index equivalent. This happens transparently and cannot be seen in the user interface.

• Where a property prediction passes through a process unit without changing (that is, the base value for the driver and prediction is 0 and the delta is 1, so that the output property value is always equal to the input property value) the property is swapped to its blend indexed equivalent with an equivalent linear pass through. So if the visbreaker gas oil pour point was determined to always be equal to the feed pour point measured in Celsius, then the pour point property with a delta of 1 would be swapped for optimization to pour point index with a delta of 1. This happens transparently and cannot be seen in the user interface.

Where this is not possible, AVEVA Unified Supply Chain must introduce a dynamic conversion between engineering space and blend index space. This involves introducing the blend index formula into the optimization problem. As the blend index formula is likely to be non-linear, this introduces non-linear structure into the optimizer, which is best avoided.

For example, the visbreaker pitch viscosity may be 0.25 times the feed viscosity (measured in cSt). If this value is placed into the base delta structure as is, it is necessary to introduce the blend rule into the underlying optimization equation. When this occurs you will receive the diagnostic warning Process unit requires non-linear conversion to blend rule. This message indicates that a non-linear element has been automatically introduced into the problem.

It may be possible to reformulate the base delta equation to drive and predict the property value in blend index space originally. That is, rather than expressing the driver, prediction and delta values in cSt their non-linear blend index equivalent are used. For example, if refutas was used as the viscosity blend rule, expressing the feed viscosity, pitch viscosity and delta relationship in refutas would mean that the structure was already linear. This would avoid introducing the non-linear conversion structure (Predictions using Blend Rules).

True boiling point blending in Plan

Distillation properties such as True Boiling Point (TBP) are typically used as product specifications in refinery models. However, there are some subtleties with the way in which these properties are blended.

As an example, consider some butane being blended into a gasoline pool, and in particular the 95% point of the boiling curve.

Butane is a single component that all boils at one temperature, which results in the following boiling curve:

The 95% point is where the dotted 95% line crosses the boiling curve, around –12°C.

The rest of the gasoline pool has a boiling curve like the following:

Here the 95% point is around 208°C.

By blending the property linearly, if we blended 10% butane into the remaining 90% of the gasoline pool we would get a 95% point for the blend as follows:

However, if we blend the curves together and measure the curve for the blend, we can see that this is not the case:

The resultant 95% point is around 205°C. Clearly, some more subtlety has to be introduced in how this value is calculated.

Note that this is a deliberately extreme example, and for many uses this linear blending approach may be sufficiently accurate.

Implementation in Plan

Whilst it is not possible to blend the fixed percentage properties linearly (such as the 95% point, measured by horizontal lines on the plots), it is possible to blend the fixed temperature points (such as the percentage remaining at 100°C, measured by vertical lines on the plots).

In AVEVA Unified Supply Chain the property for the fixed percentage values is called True Boling Point of X%. For example, above the 95% point it would be True Boiling point of 95%. The property for fixed temperature values is called Cumulative yield at X°C/F. For example, the percentage remaining at 100°C would be Cumulative yield at 100°C.

It is also possible to infer the true boiling point values from the cumulative yield values. For instance, in the blend example above, if we know that the blend cumulative yield at 200°C is 92.8 and the cumulative yield at 210°C is 96.4, the true boiling point of 95% must be between 200°C and 210°C. We can linearly interpolate between these points, and estimate that the TBP of 95% is 206.1°C.

Plan can do this interpolation automatically to avoid blending these TBP values linearly. It does this by the following steps:

1. Linearly blend the cumulative yield values of the blend components, to determine the cumulative yield values of the product.

2. Linearly interpolate between these cumulative yield values to calculate the requested TBP values, and ensure they fall within the product’s specifications.

Configuration

There is no configuration option to activate this feature in Plan. Instead it is invoked automatically when the following criteria are met:

• A TBP specification is provided on a product.

• Multiple cumulative yield predictions are provided on all of the blend component streams.

• The same cumulative yield temperatures are specified on all component streams.

• The required TBP property is not predicted on all component streams.

• A different correlation for True Boiling Point is not provided on the Correlations page.

  Note: TBP values are determined by a linear interpolation around the provided cumulative yield points. When building a model, ensure that the cumulative yield points predicted are sufficiently dense in the region of interest of the boiling curve (that is, where the product specification is likely to be).

  Below is a table with some situations to help clarify in what cases this feature is used:

  Component 1 properties predicted	Component 2 properties predicted	Product specifications	TBP interpolation used?
  Cumulative yields at 100, 150, 200, 250°C.	Cumulative yields at 100, 150, 200, 250°C.	TBP of 95%.	Yes.
  Cumulative yields at 100, 150, 200, 250°C.	Cumulative yields at 110, 155, 200, 24°5C.	TBP of 95%.	No: error returned.
  Cumulative yields at 100, 150, 200, 250°C; TBP of 95%.	Cumulative yields at 100, 150, 200, 250°C; TBP of 95%.	TBP of 95%.	No: component TBP values blended linearly.
  Cumulative yield at 200°C.	Cumulative yield at 200°C.	TBP of 95%.	No: error returned.
  Cumulative yields at 100, 150, 200, 250°C.	Cumulative yields at 100, 150, 200, 250°C.	Cumulative yield at 200°C.	No: no TBP value calculated.