Digital Copper Corrosion Measurement Vs. Visual Rating _ Incorporating New Technologies To Method Development

Transcription

1 Digital Copper Corrosion Measurement Vs. Visual _ Incorporating New Technologies To Method Development Aaron Mendez Ph.D. and Juan Ayala Ayalytical Instruments Inc W Fulton St, Chicago, IL USA Galveston, TX October 16-17, 2018

2 ASTM D130 Corrosiveness to Copper from Petroleum Products by Copper Strip Test A polished copper strip is immersed in a specific volume of the sample being tested and heated under conditions of temperature and time that are specific to the class of material being tested. At the end of the heating period, the copper strip is removed, washed and the color and tarnish level assessed against the ASTM Copper Strip Corrosion Standard. Claret Red * Magenta overcast on brassy strip Slight Tarnish Moderate Tarnish Dark Tarnish Corrosion Multicolor Lavender or silver or both overlaid on claret red * Multicolor red and green showing peacock, but not gray

3 Reasons for Development One of the last Remove BIAS from Operator to Operator Digital Recording of Images LIMS Integration Standardization on Light Source Magnification of Sample Strip Strip Size Determination Current Chart NOT REAL Fixed corrosion criteria Automatic Measurement Not assessment More reliable Repeatable and Reproducible A better rating

4 THE DESIGN Pogo Pin Connectors Spring Loaded Strip Holder NFX Strip Holder* Digital High Resolution CCD Camera D130 Copper Test Strip CuDDI Analyzer* *Patent Pending

5 Vision Algorithm Define Strip Area Automatically Inspection Area Determination 8x8 Pixel scan, ~96,000 points total Accumulate Results Calculate Percentage of each Pass values through Support Vector Machine and Algorithm for Categorizing Ability to Train and Adopt Model

6 D130 Critical Points Lighting Fingerprints Dimensions Proper Polishing Holes in D1838 Water Spots LPG Samples 2a to 3a Logic Handling 2c to 3b Logic Handling

7 WK62219 Revision of D Revision of D Standard Test Method for Corrosiveness to Copper from Petroleum Products by Copper Strip Test Rationale: D is the standard method for the determination of the degree of corrosiveness of petroleum products with a vapor pressure up to 124kPa (37.8 C), such as natural, automotive and aviation gasoline, aviation turbine fuel, kerosene, diesel fuel, distillate fuel oil, cleaners solvent (Stoddard), lubricant oils and other hydrocarbon fractions falling within the specified vapor pressure limit. A polished copper strip is immersed and heated for a determined time in a predetermined volume of the sample under test. Heating conditions and time of exposure are specific of the class of material being tested. After the heating, the strip is removed washed and the degree of corrosiveness of the sample is assessed by visual comparison of the level of tarnish and color against an appropriate ASTM Copper Strip Corrosion Standard. The proposed revision improves the rating process by a Digital Detection Imaging analyzer removing the subjective visual rating of D130, strictly adhering the classification specified in Table 1. Exact levels of corrosion are digitally and integrally identified in a 4-step automated vision algorithm process using a highresolution CCD camera with optical intelligence.

8 WK #62219 Revision of D BALLOT ITEM D02.05 (18-04) Date: 05/07/2018 To: Subcommittee D on Properties of Fuels, Petroleum Coke and Carbon Material Tech Contact: Aaron Mendez, Work Item #: Ballot Action: Test Revision of D130-18/ Standard Test Method for Corrosiveness to Copper from Petroleum Products by Copper Strip Rationale: ASTM D130 is the standard method for the determination of the degree of corrosiveness of petroleum products that exhibit a vapor pressure up to 124kPa (37.8 C). A polished copper strip is immersed and heated for a determined time in a predetermined volume of the sample being tested. After the heating period the strip is removed washed and the degree of corrosiveness of the sample is assessed manually by visual comparison of the level of tarnish and color against an appropriate ASTM Copper Strip Corrosion Standard. This ballot proposes to add a new automated rating procedure of the copper strip, using a high-resolution CCD camera with optical intelligence and digital image analysis technique.

9 WK #62219 Revision of D BALLOT ITEM D02.05 (18-04) 1. Most product specifications that have a copper strip corrosion requirement set the limit at No. 1 maximum. So the critical rating issue is whether the copper strip is 1a / 1b or if there is any hint of rose color for No. 2 or higher rating. Whether the strip is No. 2b or higher is not very relevant it is a failure. An ILS should concentrate on validating the new optical reader versus visual rating at the No. 1b to No. 2a level. 2. While D975, the diesel fuel specification, allows a No. 3 maximum copper strip rating, I doubt that many ULSD fuels exceed No. 1b. 3. At this point, I would not recommend the automatic reader as there has not be a full ILS with published precision and bias data.

10 WK #62219 Revision of D BALLOT ITEM D02.05 (18-04) 4. When adding a second tube rating technique, the Precision and Bias is critical to ensure consistency in test results. In this case, the ballot is providing an alternative solution to reading the test coupons; visual reading vs automatic reading done with an algorithm. A full ILS in accordance with ASTM D6708, Standard Practice for Statistical Assessment and Improvement of Expected Agreement Between Two Test Methods that Purport to Measure the Same Property of a Material, needs to be completed.

11 EXPERIMENTAL RESULTS Display of results and rating in Automatic mode

12 EXPERIMENTAL RESULTS Display of results and rating in Automatic mode

13 EXPERIMENTAL RESULTS Internal Repeatability Study: A compiled data set of the 13 Copper strip samples run in three instruments by three operators is presented as Table 2 below: Manual Referee rating=> 1a 1a 1b 1b 2b 2b 2a 3a 2d 2d 3a 4a 4a RIG # Operator coup1 coup2 coup3 coup4 coup5 coup6 coup7 coup8 coup9 coup10 coup11 coup12 coup13 Instrument1 Oper.1 1a 1a 1b 1b 2b 2b 2a 3a 2d 2d 3a 4a 4a Instrument1 Oper.2 1a 1a 1b 1b 2b 2b 2a 3a 2d 2d 3a 4a 4a Instrument1 Oper.3 Instrument2 Oper.1 1a 1a 1b 1b 2b 2b 2a 3a 2d 2d 3a 4a 4a Instrument2 Oper.2 1a 1a 1b 1b 2b 2b 2a 3a 2d 2d 3a 4a 4a Instrument2 Oper.3 Instrument3 Oper.1 1a 1a 1b 1b 2b 2b 2a 3a 2d 2d 3a 4a 4a Instrument3 Oper.2 1a 1a 1b 1b 2b 2b 2a 3a 2d 2d 3a 4a 4a Instrument3 Oper.3 Table 2. Internal comparative data

14 EXPERIMENTAL RESULTS Sample ID D130 Visual CuDDI 1 1b 1b 2 3a 2b 3 2a 2a (Highest) 3b (Most Corrosive) 4 1a 1a 5 2e 2e (Highest) 4a (Most Corrosive) 6 4c 4c 7 2b 4a 8 1b 1b (Highest) 2a (Most Corrosive) 9 2c 2c 10 4a 4a 11 3b 3b 12 4c 4a 13 2a 2a 14 2c 2c (Highest) 4a (Most Corrosive) Table 3. Internal Correlative Study

15 EXPERIMENTAL RESULTS Combined Stream Date D130 Visual CuDDI Automatic 5/9/2018 1B 1A 5/10/2018 1B 1A 5/11/2018 1A 1A 5/12/2018 1B 1A 5/13/2018 1A 1A 5/14/2018 1B 1B 5/15/2018 1B 1B 5/16/2018 1B 1A 5/17/2018 1B 1B 5/18/2018 1A 1A 5/19/2018 1A 1B 5/20/2018 1A 1A 5/26/2018 1A 1A 5/26/2018 1A 1A 5/26/2018 1A 1A 5/29/2018 1A 1A 6/2/2018 1A 1A 6/4/2018 1B 1B 6/5/2018 1A 1A 6/5/2018 1A 1A 6/6/2018 1B 1B 6/7/2018 1B 1B 6/9/2018 1B 1A 6/10/2018 1B 1A 6/13/2018 1B 1A 6/19/2018 1A 1A 6/21/2018 1A 1A Table 7. External CuDDI vs. D130 Data Combined Stream * * * Date D130 Visual CuDDI Automatic 5/8/2018 2C 4A 6/5/2018 3A 2D 6/12/2018 1A 2C 6/3/2018 1B 2A

16 EXPERIMENTAL RESULTS Table 6. External CuDDI vs. D130 Data Sample Date D130 Visual CuDDI Automatic Mid Stream 5/9/2018 1B 1A 5/10/2018 1B 2A 5/11/2018 1B 2A 5/20/2018 1B 1A 5/29/2018 1B 1B Sample Date D130 Visual CuDDI Automatic Mid Stream 5/22/2018 3B 4A 6/6/2018 3B 4A 6/4/2018 3B 4A 6/5/2018 3B 4A 6/7/2018 3B 4A 6/10/2018 3B 4A 6/20/2018 4A 4A Sample Date D130 Visual CuDDI Automatic Mid Stream 5/5/2018 2E 2E 5/8/2018 2B 2B 5/15/2018 2B 2A 5/16/2018 2A 2B 5/18/2018 2A 2C 5/26/2018 2C 2C 5/27/2018 2B 2B 5/12/2018 3B 2B 5/13/2018 3B 2C 5/14/2018 3B 2C 5/17/2018 3B 2C 5/29/2018 2B 2B 5/29/2018 2A 2B 6/2/2018 2C 2C 6/4/2018 2A 2A 6/9/2018 2C 2D 5/29/2018 2B 2B 5/29/2018 2A 2B 6/2/2018 2C 2C 6/4/2018 2A 2A 6/12/2018 2C 2D 6/13/2018 2C 2C 6/19/2018 2D 2D 6/21/2018 2D 2D 6/8/2018 3B 2C

17 EXPERIMENTAL RESULTS Table 4. External CuDDI vs. D130 Data Sample Date D130 Visual CuDDI Automatic Feed # 1 5/10/2018 2A 2A 5/11/2018 3B 2A 5/12/2018 1B 2A 5/14/2018 1B 2A 5/15/2018 1B 2A 6/3/2018 1B 2A 6/4/2018 1B 2B 6/9/2018 1B 2A 6/3/2018 1B 2D 5/5/2018 2C 2C 6/4/2018 1B 2D Sample Date D130 Visual CuDDI Automatic Feed # 2 5/10/2018 2A 2A 5/11/2018 3B 2A 5/12/2018 1B 2A 5/14/2018 1B 2A 5/15/2018 1B 2A 6/3/2018 1B 2A 6/4/2018 1B 2B

18 EXPERIMENTAL RESULTS Table 5. External CuDDI vs. D130 Data Sample Date D130 Visual CuDDI Automatic Feed # 3 5/9/2018 3B 3A 5/10/2018 3B 3B 5/11/2018 3B 3B 5/13/2018 2C 3B 6/17/2018 3A 3A Sample Date D130 Visual CuDDI Automatic Feed # 4 5/4/2018 3B 4A 5/8/2018 3C 4A 5/13/2018 3B 4A 5/16/2018 3B 4A 5/17/2018 4A 4B 5/18/2018 3A 4A 5/20/2018 4A 4A 5/26/2018 4A 4A 5/27/2018 4A 4A 5/29/2018 2B 4A 6/5/2018 3B 4A 6/5/2018 2C 4A 6/6/2018 2C 4A 6/7/2018 2C 4A 6/8/2018 3B 4A 6/8/2018 4A 4A 6/10/2018 4A 4A

19 Table X. Product span for D ASTM Standard Designation Standard Specifications for: D Fuel Oils D Diesel Fuel Oils D a Aviation Turbine Fuels D Gas Turbine Fuel Oils D Denatured Fuel Ethanol for blending with gasoline for use as Automotive Spark-Ignition Engine Fuels D Automotive Spark Ignition Engine Fuels D D a Ethanol Fuel Blends for Flexible-Fuel Automotive Spark-Ignition Engines Jet B Wide-Cut Aviation Turbine Fuel D ce1 Biodiesel Fuel Blend Stock (B100) for Middle Distillate Fuels D Diesel Fuel Oils & Biodiesel Blends (B6 to B20) D Aviation Turbine Fuels containing Synthesized Hydrocarbons D D Unleaded Aviation Gasoline Test Fuels containing non-hydrocarbon components Natural Gasoline as a Blend Stock in Ethanol Fuel Blends or as a Denaturant for Fuel Ethanol D RON Test Fuels for Automotive Spark-Ignition Engines

20 D Precision Statement Results by this test method are nonnumerical and involve multi-categoric rating classifications therefore conventional D02 statistical techniques, such as Practice D6300, are not suitable for determining precision. Instead, a statistical simulation approach was used to mathematically assess the degree of disagreement under r and R conditions for the ILS data set. Based on the statistical approach followed, no statistically significant difference in disagreement percentages between r and R conditions was determined for rating classifications in category 1, 2, and 3. For rating classification 4 however, samples showed more withinclassification disagreement. Based on this information, the statistician determined that in the long run, the precision of the analysis (that is, both r and R ) is that there is nominally a 5% chance that a difference between any two ratings will exceed the criteria in D130 Table 2.

21 D Revision Plan D b _ Standard Practice for Statistical Assessment and Improvement of Expected Agreement Between Two Test Methods that Purport to Measure the Same Property of a Material The interlaboratory study must be conducted on at least ten materials that span the intersecting scopes (See Table X) of the test methods and results must be obtained from at least six laboratories using each method. Similar cases were automatic procedures with its own precision were added to the standard, specifically D _ Smoke Point and D3241_ Thermal Oxidation Stability of Jet Fuels (JFTOT). A common set of representative coupons previously rated according to D130 can be circulated among cooperators to determine the automatic rating. The minimum number of samples and the degrees of freedom involved is something that must be determined with ASTM D02.94 statisticians.

22 CONCLUSIONS CuDDI possesses a tight controlled analyzer that guarantees high consistency High resolution CCD camera provides precise machining and motors to reproducibly rotate the strip The novel Pongo Holder eliminates unwanted markings that might affect the measurement s proceed by accurately measuring the degree of corrosion eliminating operator bias in the visual subjective assessment Low data dispersion between tests and operators/instruments CuDDI presents unique advantages in avoiding distortions around fringing areas and in determining the real copper strip size

23 CONCLUSIONS In the case of pass/fail data, no generally accepted method for determining precision is currently available, except for gasoline sample types (see D130-18, and ) analyzed using the test tube procedure stated in D Bias has not been determined for this test method. A full fledged Round Robin Study covering all types of fuels is being prepared before resubmitting this work item for approval.

24 THANK YOU! QUESTIONS?