UOP 7238A-0 © 2016 UOP LLC. A Honeywell Company All rights reserved.

James Rekoske

February 7th, 2017 CARB Meeting Sacramento, CA

Mass Balance Approach for Co-processing

UOP Experience with Unit Mass Balances

• UOP uses mass balances for daily unit monitoring, troubleshooting, and performance testing of FCC and DHT Units- Challenges include occasional error in the inputs which can have an

impact on calculated outputs- Data collection frequency and duration can also have an impact on the

accuracy of the material balance - FCC units typically have a material and energy balance conducted for a

single day of operation, using laboratory data collected during a shift coupled with daily averages of process data

• Feed, process, and product streams flowrates on the unit are measured and are largest contributor to the quality of the mass balance

1

Definition of a Good Mass Balance

• In UOP’s experience good unit mass balances typically consist of:- Mass Balance – 98-102 wt% - Confirms flow meters are correct- Hydrogen Balance – 98-102 wt% - Confirms lab data is correct (i.e.

Gravity and Distillations)- If out of range, normal troubleshooting efforts will be undertaken to

understand source of variation

2

51464136312621161161

105.0

102.5

100.0

97.5

95.0

Observation

Indi

vidu

al V

alue

_X=98.82

UCL=100.74

LCL=96.91

51464136312621161161

3

2

1

0

Observation

Mov

ing

Ran

ge

__MR=0.721

UCL=2.356

LCL=0

11

I-MR Chart of Mass Balance

Analyzing Commercial Co-processing Trial Mass Balance Data for Biocrude Feed Impacts

• Key- Long term testing is needed to accurately measure small shifts in yields - Conversion varies about 0.7 to 1.0 v% due to natural variation in

measurements, catalyst activity, feed quality, cut points, etc- Because of this natural variation, many observations are needed to

measure small yield shifts to desired accuracy

• Solution- Use simple models to adjust the raw data

3

Proposed Approach for Measuring Biocrude FCC Co-processing Performance

• Measure- Baseline performance based on current or historical data without biocrude

• Model- Develop simple statistical regression model of yield relationships to

conversion (and other parameters) with small and defined error bands based on observed commercial performance

• Test- Complete both pulse and long duration testing with biocrude to generate

yield performance data with biocrude

• Tune- Incorporate a biocrude regression term into the original yield model and

regress equations based on observed performance from test

• Predict- Use the model to predict performance with and without biocrude with current

data to calculate yields attributable to the presence biocrude

4

Measure Model Test Tune Predict

14

15

16

17

18

19

20

21

22

-0.01 0.00 0.01 0.02 0.03 0.04 0.05 0.06Bio-oil Fraction Leverage, P=0.0043

Hypothetical Example of Model Output5

Gasoline Yield LCO Yield

44

45

46

47

48

49

50

51

-0.01 0.00 0.01 0.02 0.03 0.04 0.05 0.06Bio-oil Fraction Leverage, P=0.7501

6

UOP 7238A-6

Establishing Statistically Significant Trends• Regression Example

- Mass balance / yield data can be worked up to establish clear trends as a function of biocrudeaddition

- Proposed simple linear model of key variables over modest ranges- Responses are “trained” on pre-test, post-test daily operations by simple regression- Fraction of Biocrude Feed: (FBF) is in each model to quantify yield shift (if any) and measure

significance

• Possible Equation form- Conversion – FBF, Triser, Cat/Oil, Activity, VGOAPI, VGONitrogen, Throughput

7

Typical FCC Yield Factors & Relationships

• Conversion – f(FBF, Triser, Cat/Oil, Activity, VGOAPI, VGONitrogen, Throughput)• C2- = f(FBF, Conversion, contact time, metals)• LPG = f(FBF, Conversion, contact time, ZSM5, metals, PartialPressure)• Gasoline & LCO = f(FBF, Conversion, contact time, ZSM5, metals, PartialPressure,

T90)• Coke – same variables, TFeed, TRX, CatCooler Duty

• Delta Coke – FBF, CCR/Cat/Oil, Cat/Oil, Tfeed, API, PartialPressure, Steam/Cat, Throughput (Flux, B/D)

• Cat Activity – Cat addition rate, metals (Na, V, Ca, K), Regen Temp• COX – FBF, Cat/Oil, RegenPressure, TRegen, FlueGasCO

8

Typical HYT Mass Balance

3212892572251931611299765331

120

110

100

90

Observation

Indi

vidu

al V

alue

_X=99.41UCL=101.72

LCL=97.10

3212892572251931611299765331

20

15

10

5

0

Observation

Mov

ing

Ran

ge

__MR=0.87UCL=2.84

LCL=0

1

1

1

11

1

1

11

1

1111

111

111

1

11

11

11

11

11

I-MR Chart of HY Mass Balance

9

Establishing Performance from a Co-processing Trial

• Preparation- Establish baseline mass balance / performance values for a minimum of

a month before the co-processing trial It may be necessary to recalibrate instrumentation and work with the

laboratory to ensure the analytical methods are being executed properly Data should be collected prior to the trial for a time equal to, or greater than

the proposed trial duration to best understand how the unit runs with and without biocrude and to provide basis for statistical regression and modelling

• Execution- Focus during the test will be on establishing trends and the impact that

biocrude had on product yields during the test For most commercial operations, the biocrude will only be added at 1-3 vol% The raw yield and operational shifts measured will establish gross effects

from co-processing, however it is understood that variations in feed, operating conditions, and catalyst that exist as part of normal operation can impact the results

10

How to measure the yield change in FCC11

Collect existing data (could be hundreds of data points) on the FCC to establish relationships of product yields to conversion with small and defined error bands versus conversion

7472706866646260

51

50

49

48

47

46

45

44

S 1.04872R-Sq 66.7%R-Sq(adj) 59.3%

Conversion 3% RFO

Gaso

line(

C5-2

21°

C) 3

% R

FO

Regression67% CI

Gasoline(C5-221° C) 3% RFO = 25.57 + 0.309 Conversion 3% RFO+ 0.00014 Conversion 3% RFO**2

In FCC conversion is the dominant variable affecting yields as shown below with example confidence ranges

Estimate number of biocrude tests neededInputs: Estimated BC yield to fuels

Petroleum yield confidence intervalRequired minimum fuel yield of BCRequired confidence level