RESEARCH AND DEVELOPMENT PROJECT.

(Technical report - current status)

IAMSL and AVS are undertaking a research project to gain professional third party accreditation for our system. The project is being undertaken by Jeff Foley (IAMSL).

Within the scope of the project, we have completed our work on regressional analysis of the correlations:
Pilodyn & Hardness (two-way), Hardness & Density (two way), Hardness to MoR, and MoR to E, to f'c, to ft, and to fs.

We have used all of the published information available to us on the characteristic properties of each timber species listed in AS1720.2, with two points per species, one corresponding to a moisture content of 12% (Dry) and the other to 28% (Green).
The Pilodyn - Hardness relationship is modelled by a derived formula based on standard engineering principles - we plan to validated and verify this with simple tests on sample specimens.

Under the TMXScan system, we propose using a Pilodyn or Density measurement to estimate the Hardness, and then MoR, and then other properties.

With all of the different trees (softwoods and hardwoods, 12%mc and 28%mc) represented in the regression models, we have moved away from the de-rating notion of an S2 pole (say) to an S3 or S4, but instead, take the Pilodyn (or density) measurement as an indication of how much the original timber has degraded due to aging in its particular service environment.

We assume only that the characteristic strength values published in AS1720, AS 2878, etc are obtained using consistent measurement procedures, and are not so concerned that they may be 5 percentile values, or averages, or whatever. We get around this for pole timbers by applying your NDE test results as follows: To estimate MoR, we look at two methods.
1). We take your outer timber density measurement -> Hardness -> MoR
2). We take your groundline zone failure Bending moment with Z(groundline) and Z(groundline - 300mm) in the loaded direction.
Study of the regression between the two gives us a linear equation which we use to "calibrate the system" - we take the calibrated (adjusted) MoR and say that "the pole will fail somewhere along its length when the corresponding bending moment is developed over its groundline zone". With the calibrated MoR we estimate E, f'c, ft and fs. We have found that the correlation using Z(GL-300) has a tighter regression and propose to adopt this - the pole section modulus at GL-300 better represents the residual condition (and is generally somewhat less than at GL).

We have found very good, well defined correlations between TMXScan plots and section x-cuts (about 650) for residual area, location of centroids, and 2nd moments of area about the centroids. We have also developed a very reliable method of taking a scan (based on the +150 x-cut) and projecting down 300mm (to the -150 x-cut), and hence, from groundline to GL-300mm (used in the above study). This involves some amplification and some truncation of the various anomalous segments detected - we are working on image and graphic presentation of this on a single A4 page per NDE pole.

At the conclusion of the project, we intend to present our TPXScan system of:
Pilodyn, GL Scan, GL-300 residual diameter, GL-300 Section Plot
Estimation of residual properties (MoR, E, f'c, ft and fs), Failure GL-300 Bending Moment, and Tip Load
each defined in terms of mean accuracy and standard deviation, modelled by log-normal functions, directly applicable to assessment of probability of failure (and its complement, reliability), local linear degradation rate, and remaining life.

With the acousto-pulse method of probing a x-section for areas of degradation, we have always had a problem with shadows (areas of sound wood tagged as anomalous due to being in the shadow of real cracks and pockets of rot) - though these shadows produce tolerable levels of conservatism in residual strength assessments, they make for plots that do not always ideally resemble the actual x-cut and this has sometimes caused a degree of scepticism from the novice observer.

First, we tried a simple approach of weighting the anomalous areas according to their appearance in the 650+ NDE x-cuts. We extended this to incorporate probability maps of the various possibilities, and have recently decided to examine patterns of areas of anomaly in context. The scanning system is symmetrical to the extent that there are 640 possible arrangements in each of 8 sectors of the cross section (i.e., 650 NDE x-cuts by 8 sectors giving a very large sample size of 5,200). With some clever manipulation and filtering algorithms, we are now in the process of assembling some 50 lookup tables. This has added major complexity to the TPXScan software and at least another 6 weeks onto the project, but the preliminary results are plots that are spectacular in their fidelity and quality.

BACK