CASE STUDY

EXTRUSION QUALITY MONITORING

Background One area in which particular interest has been shown by cable makers is the issue of monitoring the quality of the extrusion process - in particular the monitoring of the overall stability of the extrusion of the insulation layer. Now as UltraScreen samples each of its 16 channels every 16ms, it collects dimensional data that is ideally suited to addressing such extrusion quality issues. UltraScreen actually collects such data in data files that hold about 1500 samples per channel and thus equates to about 25s of run time, or about 40cm of cable length (at 1m/min line speed).

Longitudinal Thickness Variation The longitudinal variation in the thickness can be evaluated from this data, and the figure below presents the insulation layer thickness captured for one channel from one such data file, actually presented as the variation of this layer thickness around the average thickness for that file.

How UltraScreen can automatically detect insulation

thickness variations

From figure 1 above it can be seen that thickness varies by about +/ - 300micron over this length of cable, and this variation can be characterised by evaluating its Standard Deviation (S.D.) which equates to 118 micron for this channel.

A new way to assess

the quality of extrusion

Circumferential Thickness Variation Now as UltraScreen measures 16 such channels, the S.D. of the extrusion variation for each channel can be evaluated and displayed as in the figure below.

Figure 2 — Insulation Thickness S.D. values around the cable.

From this figure it is clear that that the extrusion quality of the insulation layer is not consistent around the cable, with the minimum S.D being measured at about 60micron, whilst the maximum is about 120micron. Also it is clear that this variation in S.D. value is not random, but has a clear pattern indicating, in this case, that the cable sector, spanned by channels 5 – 7, has an extrusion stability that is worse than any other sector.

This sector can be related back to the actual geometry of the extruder head to identify the aspect of the extruder head that is responsible for the decrease in extrusion stability.

Using Fourier Transforms to identify the frequency of variation

Frequency Analysis The time series shown in Fig 1 can be analysed, using Fourier Transform (FT), to identify the frequency characteristics of this extrusion variation, and the analysis of the data set is presented in Fig 3. This shows the frequency characteristics of the data in the frequency band up to 4Hz, with this data scaled to accurately reflect the 16ms sampling period fundamental to the UltraScreen time series data collection.

Fig 3 - Insulation Layer Thickness Variation, Frequency Analysis (0 – 4Hz)

From Fig 3 it may be seen that there is a large peak (blue arrow) in this data, at the left hand side (low frequency end) of the FT ‘Spectrum’ presented in the figure. The frequency of this peak is at ~=80mHz which implies a Periodic Time (PT) for this oscillation of ~12.5s, but as there is also quite a large response also at ~= 40mHz, this FT response suggests that the ‘mechanism’ producing this oscillation in the layer extrusion probably has a PT a bit longer than this, so probably more like ~15 - 20s.

Identifying the causes of variation

Summary This note has explored the potential use of UltraScreen to monitor the quality of the extrusion process, by providing quantitative measures and displays of data covering:

Longitudinal layer width variation. Circumferential layer width variation. Frequency analysis

The feedback provided by this information has the potential to not only identify the sector of the extruder head responsible for increased extrusion instabilities, but also the frequency characteristics - and hence the nature - of the extrusion mechanisms causing the instabilities. All of this processing is only possible because of the rapid sampling regime executed uniquely by UltraScreen. This underpins the detailed assessment of both the physical and the frequency characteristics of the layer extrusions presented in this case study.

Also, there would seem to be a second peak (red arrow) with a peak frequency ~=330mHz, PT ~ 3s. This again would suggest the presence of a second mechanism that is producing an oscilla-tion in the extruded layer width with a shorter periodic time. So the analysis of this data set suggests that there are two mechanisms producing oscillations in the insulation layer thickness.

A lower frequency mechanism, PT ~= 15 - 20s. A higher frequency mechanism, PT ~= 3s.

Thus such frequency analysis opens up the possibility of using such analysis as a ‘diagnostic’ process that identifies the frequency characteristic of the mechanisms within the extrusion process responsible for the instability seen in the layer extrusions. With this new knowledge this potentially allows the mechanisms themselves to be identified and improved.

TECHNOLOGY- BASIC PRINCIPLES

LEFT :

16 channels are shown in 2D

format which represents the

total 360 degree view of the

cable under evaluation

Single channel representation

BELOW: Energy from the transducer is

transmitted to the cable via the water. Energy is

reflected back to each single transmitter /

receiver at every boundary where the acoustic

impedance is not matched.

A RING OF 16 ACOUSTIC TRANSDUCERS

SURROUNDS THE CABLE AS IT PASSES

SLOWLY THROUGH A WATER TANK

INSIDE ULTRASCREEN.

Cable is scanned through 360 degrees

every 16mS. This equates to a full scan of

cable parameters and features every 250

microns (at 1 m / minute typical line

speed).