Proceedings of KGCOE-MD2004: Multi-Disciplinary Engineering Design Conference Page 5

MERIT 05100

Non- Destructive Joint Test Method & Equipment Development

Aleksandra Solda -- Project Manager/ISE
Joshua Chan -- ME / Elizabeth Kesel -- ME / Robert Lefferts -- ME
Thinh Cao -- EE / Jason Mitchell -- EE / Lauren Williams -- ISE
James Taylor – Faculty Mentor

Proceedings of KGCOE-MD2004: Multi-Disciplinary Engineering Design Conference Page 5

Abstract

The technical paper presented will summarize the equipment selection process, and experimental progress completed by Senior Design Team 05100 in regards to verifying the long-term integrity of the clinch portion of a heater core pipe to neck shell. The main goal of the project is to develop an accurate way to measure the clinch on a heater core assembly for an assembly process at Delphi Thermal & Interiors as well as to understand a correlation between joint integrity and the pressures and temperatures that cause leak.

introduction

The particular heater core is an air to liquid, aluminum heat exchanger used in a HVAC module for automotive applications. Coolant from an engine is passed through the heater core to provide heat for the interior of a vehicle.

Delphi Thermal & Interiors has been producing this particular type of heater core for over a decade. The clinching method has also been used for the past decade to connect inlet and outlet pipes to the heater cores themselves.

A clinched joint is defined as a pipe with an o-ring that is inserted into the neck shell of a heater core. This assembly is then entered into a fixture to hold the core as the jaws on a cam plate roll to deform the neck shell onto the bead of the pipe. The deformation of the metal causes a gap between the pipe’s outer diameter and the clinch. A maximum gap size of 2 mm is allowed as shown in Figure 1.

Fig. 1

Delphi would like the team to develop a method to measure the clinch on each heater core in a production environment. This method will be used to verify that the clinch on every part meets specification.

Delphi would also like to see if there is a correlation between the clinch size and leak rate. We will perform experimental testing in a lab environment to try to understand the relationship between leak rates and clinch gap. There is a maximum leak rate of 6 cubic centimeters of air per minute for each heater cores.

A selected group of heater cores will undergo leak testing at specific temperatures and pressures. This testing will help us determine whether or not there is a correlation between clinch gap and leak rate.

Figures 2 and Figure 3 exemplify the heater core and a clinch of a neck shell respectively.

Figure 2

Figure 3

Leak Testing Setup

The heater cores were tested with air instead of water as a result of test equipment availability. Testing was performed at Delphi Energy & Chassis located in Henrietta, NY. Leak measurement fixtures were modified to meet the testing needs. The maximum leak specification of 6 ccm applies to testing the units with air.

Several test plans were discussed to choose the most efficient test to meet the objective of the project. The proposed test plan from the RIT design team called for a sample size of 30 parts; each part would be tested at five different temperatures and four different pressures. The temperatures and pressures were selected based on given operating conditions of a thermostat in a vehicle. These temperatures range from 195 º F – 240 ºF. Figure 4 shows a detailed structure of how the data will be taken.

Figure 4

Another test plan considered was a plan similar to a DOE. This was suggested because there were concerns about the heater cores having a memory effect which would affect the leak rate at the end of the test. The heater cores would be grouped into one of three different clinch ranges. Parts would be randomized so that each heater core would only be subjected to one pressure/temperature scenario. Each of the sixteen cells would have five parts. We would run this plan once for each of the three clinch gap ranges. Figure 5 shows a matrix of this test plan.

Figure 5

After much thought and discussion amongst the design group and with Delphi, a test plan was decided on that was similar to the first. This test plan would better meet the goal of understanding the minimum stress needed to induce a failure. A sample size of 70 parts will be used and a new temperature will be introduced. Each clinch was measured with pin gauges. There are six measurements per heater core. The test plan consisted of two groups of cores, good cores and bad cores. Good cores consisted of heater cores with clinches all less than or equal to 2 mm. Bad cores consisted of heater cores with all clinches greater than 2mm. A gauge R&R was performed on the measurement system to verify that it was reliable. Figure 6 shows a structure of how the data will be taken.

Figure 6

The next major obstacle in testing was the test setup. The heater cores were put into an oven to reach the desired temperature for each test. A mass flow meter was used to measure leak out of the heater cores. Figure 7 shows a schematic of how each heater core was physically tested. The outlet pipe is sealed with a rubber stopper and JB Weld. Connections to the inlet pipe of the heater core consisted of air supply with a pressure readout, a mass flow meter with a leak readout, and a pressure regulator.

Air flows through a pressure regulator into a mass flow meter, which has a LCD display for calculated leak rate. The air then flows into a T fitting where at one end, gauge pressure is measured and displayed and at the other end, a heater core is attached. The pressure regulator is adjusted until the gauge pressure readout displays the gauge pressure the test requires. The leak rate is recorded once the system stabilizes.

Figure 7

Camera system Development

The purpose of the measurement system is to measure the gap between the clinch and the pipe on the heater core. The factors involved with measuring the clinch gap are: repeatability, reliability, accuracy, and time. The main objective is to measure 100% of the heater cores on the assembly line at Delphi Thermal & Interiors. Because of the high production schedule, there is a very limited amount of time for an accurate measurement of the clinch gap. It is essential that the system be highly reliable to keep the production time of the heater cores to a minimum.

There were four different measurement systems suggested. The first system suggested was measuring the tool travel of the jaws that clinched the neck shell. The second system was a camera system that will take an image of the clinch and then calculate the gap width. The third system was a laser system. The lasers would reflect off of the neck shell and the width of the beam would be measured. The last method suggested was to use X-rays to obtain a clear picture of the clinch.

The system decided on was the camera system through a feasibility analysis. This option was determined to be the easiest to implement. This system has the capabilities to communicate directly with the PLC sending GO/NOGO signal to the PLC. It requires a small amount of code to be added to the PLC to look for the GO/NOGO signal from the camera system. Figure 8 shows a picture of the camera system decided on by team 05100.

Figure 8

The above figure displays a camera system produced and distributed by Keyence. The following are camera specifications for the Part #CV-2100 Series.

· Digital Image Transfer

· Repeatability of + 0.05 pixels

· Ultra high-speed processing of 20,000 parts/min

· On-screen statistical processing

· Edge area correction

· Trend edge function

· High-speed rotation search

Leak Testing Results

The test plan shown in Figure 6 was performed starting at 74 º F. Results were recorded at this initial temperature. Parts were then placed into an oven at 175 º F. Each core was removed one at a time and leak tested at 25 psi. Each heater core had a leak value of over 6 ccm at 25 psi. The team was to stop testing each core if leak had surpassed 6 ccm. The heating portion of the test was stopped and cold testing was to be performed. The cores were removed from the oven and allowed to settle to room temperature overnight. All heater cores were placed into a freezer set to – 40 º F for 1 hour. Heater cores were removed one at a time and leak tested at 25 psi. Every heater core that was tested at – 40 º F leaked 0 ccm.

Upon contacting Delphi Thermal & Interiors with our results, it was confirmed that the leak rate would indeed increase with increasing temperature. The data did show the expected changes in leak but they were very extreme because heater cores were not supposed to surpass 6ccm of leak at 175° F and 25 psi. Testing was put on hold and there was an investigation as to what could be causing these results. Potential causes of leak were discussed to be the following:

· The cores themselves were leaking

· There was too much movement in the heater core pipes causing stress in the clinch

· Internal leaks in the measurement system

· Un-calibrated equipment

· Faulty equipment

Each one of these potential causes was investigated and we determined the problem to be the mass flow meter, it was not rated for the temperatures above 104 º F. The mass flow meter used was a MKS 258c model. The meter is only rated for a temperature range between 59° F - 104° F. Delphi Thermal & Interior shipped the team an Advanced Specialty Gas Equipment Series 150k flow meter. This unit is rated up to 200° F. Figure 9 displays the set-up that was used with this flow meter.

Figure 9

In order to make this unit work a pressure decay test was setup. In a pressure decay test, heater cores are pressurized to their desired pressure. A valve is closed to trap the air within a closed system and pressure drop is monitored.

Due to time restraints we were unable to get this set up working.

Leak Testing Statistical Analysis

The team met with difficulties in designing the experiment with the most valid statistical analysis. While there were many choices in how the experiment was to proceed, few plans were feasible given time and material constraints. To demonstrate the teams understanding of engineering methods, the statistical implications of the constraints will be discussed.

The sponsor donated eighty heater cores. The best situation is for each heater core to be exposed to one treatment condition. However, this would have amounted to needing hundreds of cores. An engineer at Delphi expressed his confidence that previous conditions did not affect any future experimental results. In future testing, however, it would be statistically optimal to experiment on only one core so that the results of an experiment are due to the specific treatment conditions imposed on the core and not due to changes in the material from previous trials.

Because of time constraints and limitations of the test equipment, each core saw the exact same experimental treatments in the same order. One may say that since the experiment was run over the same sequence of events for each core, the conclusions drawn apply best to that particular series of events. Randomization allows for the effects of some nuisance variables to be assigned to error instead of being consolidated with factors of interest. For future testing, complete randomization of the experimental conditions would allow stronger conclusions about the data. One would want to choose cores randomly from a large variety and run random pressure and temperature conditions on them.

If there were more time, the experiment would have been divided evenly into sessions so that blocking could control nuisance variables associated with running the experiment on different days. In the team’s experiment, the main goal was to finish the trials in the timeliest manner possible so a performance analysis could be completed by the scheduled end of the quarter.

There are some deficiencies in the experimental procedure conducted due to resource constraints. If the customer desires future testing, the above suggestions will help lead to an even more solid statistical analysis of the heater cores.

From the data collected in the leak testing previously discussed, a statistical analysis was performed in order to find a correlation in leak to clinch gap and the changes in pressure and temperature. Summaries of the statistical findings are as follows.

When the cut off between a good and bad clinch is 2.0 or 2.5, the difference between the leak rates of good and bad clinches is between 0 and 0.3 sccm. When the cut-off is 3.0 mm, the difference between the leak rates is 0.07 and 0.4 sccm.

No particular clinch stands out as a predictor of leak rate. The average inlet pipe clinch gap is, however, a better predictor of leak rate than the outlet.

Whether or not the core has a bad clinch (as currently defined, 2.0 mm) is a poor predictor of leak rate. There is a positive relationship between median values of leak rate for each pressure and the values vary between 0.6 sccm and 1.2 sccm. Pressure and clinch gap (the average inlet pipe clinch gap) account for 16% of the variability in leak rate. Adding the interaction effect between clinch gap and pressure weakens the model slightly. Finally, the ANOVA and regression assumptions of normality are grossly violated.