Febi Bilstein leaderboard September 2020
NGK leaderboard Sept to Nov 2020
Walker Products leaderboard September 2020
Bosch leaderboard July 2020 on
House leaderboard ad for subscriptions Aug 20 on
Sealey secondary leaderboard Sept 2020
ZF secondary leaderboard Sept 2020 on
Philips secondary leaderboard September 2020

How to critically analyse vehicle data

0
In my role as a race team data analyst, I spend hours looking at data. It has been captured by various sensors on the race car and then turned into graphs or squiggly lines, as my colleagues prefer to call them, by software programs. The ability to manipulate this data into something more meaningful in terms of performance analysis is a key role for the data engineer. We use maths channels and along with the software, they can turn a basic data capture into a sophisticated analysis platform.
For example, drivers always feedback on understeer or oversteer characteristics of the car, you will see them on TV with the imaginary steering wheel turning the corner, giving feedback to the engineer. They are sometimes very wrong and this is where a simple maths channel can be used alongside steering angle input to prove how the car is actually behaving during a corner. Using the longitudinal acceleration input and dividing this by the wheelbase of the car, we get a representation of the corner radius, which can be compared to the driver’s steering input. If the driver gives a constant steering input the car is natural, if he must wind on more lock then we
have understeer and if he is winding off lock then we have oversteer. The problem is, it is a dynamic environment and you may have understeer on the entry of the corner, oversteer at the apex and natural steer on exit.
Collecting data from the steering angle sensor, while the team push the car, or drive very slowly around the corner, results in the natural steer angle for that corner, see Figure 1.
Fig1
This can be used as a comparator to make changes to the nut behind the wheel, aerodynamics, geometry or suspension to achieve the desired characteristic for the fastest way around that corner.
 INTERPRETING SPECIFIED AND ACTUAL DATA
 Data acquisition has become increasingly important in the modern world of motorsport, with limited testing allowed – reducing the carbon footprint of the sport and maximising the benefit of on-track time. The skills I use at the circuit are transferable to the workshop and the diagnostic process. We constantly gather evidence during diagnostic routines – some of it is conscious, some of it unconscious. Once we decide to look at data, we must be able to critically analyse the capture or we are wasting our time. Scrolling down through hundreds of parameters and nodding as we go isn’t particularly effective, but it is something I see all too often.
 The silver bullet of data captures is the specified value, it gives the technician a comparator, something to compare the actual reading with. If the tool can provide the data in a graphical format this is as good as it gets, but what if there is no specified value and no graph, just an endless stream of parameters all changing as the vehicle is driven along?
Many of the diagnostic tools can log data during a test drive, this means you can safely drive the vehicle, store the data and analyse it when you return to the workshop. Copy and paste the data into a spreadsheet and use the software to create graphs of specified and actual data, such as rail pressure in the  common rail engine fitted to the BMW 318d, see Figure 2.
Fig2
It is much easier using the graph to spot trends and glitches, than scrolling through thousands of lines of raw data. This is great when the tool displays the specified reading, but what if there isn’t a specified value?
 CALCULATING EXPECTED VALUES
Air Mass Meters are a typical case where unless it has failed completely, it can be very difficult to condemn the air mass meter and it is very tempting to fit a new one, just to rule it out. What if there was another way, first using live data and calculations?
We can predict air flow through the engine, it takes some practice and requires some effort, but it can be done. If you know the following, you can calculate the airflow with a high level of accuracy:
• The engine capacity
• The volumetric efficiency
• The air pressure
• The air density
• Engine speed
A simple data log will give you engine speed and manifold pressure, a quick internet search will provide air density and engine capacity if you didn’t already know it. Once you have calculated the air flow, compare your findings with the measured airflow stored in the data log, see Figure 3.
Fig3
Using the chart feature in the spreadsheet software makes this a much easier task, but collecting the data is of little use if you  are not able to interpret it. For example, a low reading from the Air Mass Meter may indicate a problem with the component or that the Exhaust Gas Recirculation valve is stuck open. It may also be a problem with the inlet manifold swirl flaps or carbon build-up. The technician has to decide the next course of action and what test to perform to prove the cause and not the effect.
 Too high a reading could mean a faulty Air Mass Meter or a leak in the intake system after the turbo. Air is passing through the Air Mass Meter but escaping before reaching the inlet manifold. The ECU will request more pressure from the turbo to compensate for the leak and a circle of doom is completed, resulting in DTC’s for the AMM. Easy to prove if you look at specified and actual manifold pressure values.
When using a scope, it is possible to collect data in the form of voltage and convert this into values using a spreadsheet. A time-consuming process, but sometimes a worthwhile one.
PicoScope users will be familiar with the custom probe function; this allows technicians to customise a given output to display not voltage but the measured value instead. This can be compared to live data or calculated data for analysis. The Diesel Particulate Differential Pressure Sensor is a good example for this. You can remove the sensor and using a pump and gauge, measure the pressure applied to the sensing element and the voltage output. Using the custom probe feature, enter this data so the scope displays the pressure not as a voltage but as Bar or PSI, whatever you prefer. This makes the analysis much easier.
Fig4
The more methods used for accurate measurement, the greater the likelihood of accurate diagnosis. If you would like to find out more about GotBoost’s training courses, visit www.gotboost.uk where you will find all the latest news and courses.

Tags: , , , , , , , , ,

Share.

About Author

Leave A Reply