Automation in the metals laboratory

Laboratories in the metal industry perform a variety of tasks that are closely tied to the production process. Examples include the analysis of samples from melting ovens, raw metals samples, water samples from the cooling circuits, exhaust gas samples, and filter dust samples, just to name a few.

In general, metal industry laboratories need to process a large number of samples in a very short time period. This and the high cost of running the factory are the main factors for automation. This automation has caused a shift from the more conventional, slow, labour-intensive, wet chemistry methods to fast and highly automated physical methods.

X-ray fluorescence and optical emission spectroscopy today are the preferred techniques. They can perform multi-element analyses of good quality in a matter of seconds. For sample preparation (eg, milling the specimen to produce a smooth surface) robots are widely used.

In such an environment, laboratory information management systems offer a number of significant advantages that increase the quality, productivity and integration of the laboratory.

From the IT perspective the picture looks like this:

• Large amount of samples (500+ samples per day)
• Fast turnaround times (< 10 minutes from sample receipt to COA)
• Samples can be planned
• Frequently only one analytical method is carried out on a sample
• No extensive validation/approval procedure required
• Automated sample registration
• Automated generation of COA and automated delivery
• High degree of integration with MRP and other systems (eg, SAP R/3, specialised production systems)

This article shows how these requirements can be met by a laboratory information management system, with a short period from initial discussions, detailed study to a productive LIMS.

Interfaces

Figure 1 shows a typical constellation detailing the co-operation of various software components in a production laboratory in the metal industry. Importantly, there are many different tasks that are tied to the laboratory work directly.

One of the central components is WinLIMS' capability to exchange data with SAP R/3.

WinLIMS receives new product specifications via the SAP R/3 interface. They are transformed into version controlled sample type test plans automatically. A software switch controls automatic approval: if on, new test plans via this route will be automatically approved and therefore can be used immediately.

Due to the large dynamic and the multiplicity of events in the production area, it is vital to make sure that every laboratory technician has the same information base at all times. As an example, laboratory personnel are informed that there is a new specification or the laboratory manager is made aware of a trend in the calibration verification runs. LIMS mail is not using the common email mechanism (which is part of the WinLIMS system as well) but rather a built-in functionality: every LIMS mail raises a dialogue that becomes the top-level window and has to be approved. This way, important information becomes available instantly, even in a noisy environment with a lot of optical distractions.

From here on WinLIMS manages the information flow to distribute the specification change which is version controlled now to several systems on the same hierarchy level.

Apart from managing specifications sample registration is an important aspect in everyday work: from the interfaces to SAP R/3 and a number of sub-systems information on production plans and corresponding sample plans is transferred into the system and used to preliminarily register sample candidates.

WinLIMS also maintains COAs from contract laboratories where certain sample types the internal laboratory cannot do themselves are being analysed. These COAs are entered manually into the SAP R/3 system and are transferred to WinLIMS via the QM-IDI interface for further use. The software stores these data in separate tables but delivers externally and internally generated data transparently to the sub-systems, so that these systems have access to all analytical information from one single source in one single format.

Laboratory life cycle

As mentioned before, a typical plant in the metal industry has a number of sources generating samples for analysis in the laboratory. This could be samples planned periodically As mentioned before, a typical plant in the metal industry has a number of sources generating samples for analysis in the laboratory. This could be samples planned periodically following a sophisticated sampling plan. This sampling plan itself can be represented in the system and can be used to generate the sample candidates automatically. Other samples are registered via different systems automatically.

A central module Sample Receipt is used for viewing the pre-registered samples and selecting those that have been actually received. This can be done by selecting rows in a table window or by scanning a barcode label on a sample. Once sample receipt has been verified the sample is automatically entered in the sample queue of an instrument for analysis.

Once more at this stage a high degree of automation is achieved with a minimum of user interaction. Based on the pre-registration of the sample the produced alloy with its specifications is already known. WinLIMS maintains a sample type for every alloy that details the methods and instruments to be used.

To be able to work with as little human interaction as possible every test method with its instrument also points to an alternate instrument and an alternate method (if required). At the time of sample receipt the database is checked for the requested instrument.

If that instrument is not available the backup instrument and backup method will be substituted automatically. This way more than 99.9 per cent of the routine work can be carried out automatically. At this stage WinLIMS also provides functionality for instrument load balancing. If a particular instrument is already under high sample load the alternate instrument will be used preferably.

Once the instrument to be used has been identified data is being transferred. As all sample information is already available no additional entries have to be made. With older instruments the traditional RS 232 style interfacing is still used, with more modern instruments communication is almost exclusively established via the ethernet using TCP/IP protocol. In these cases there are different ways for the implementation of data transfer to the instruments: a central solution handles communication from the machine with the database eg, using Oracle stored procedures.

The solution handles communication via the client computer controlling the instrument. There are advantages and disadvantages to both alternatives and the decision to go one way or the other has to be made on a case by case basis. The central approach is very maintenance friendly and safe, but the decentralised approach allows a more flexible emergency operation upon loss of the database server but requires more service.

After the analysis the laboratory technician performs a number of routine checks to verify the plausibility of the data (calibration in range, control samples in range, repeated experiments result in reasonable averages and standard deviations) and approves the data. The data is then automatically transferred to WinLIMS and entered in the database.

Because of the large number of samples and corresponding results this is always done on the DBMS side using triggers and stored procedures. In that process the data is being checked against a number of method specific limits (operating range of the method) as well as the specifications for the produced alloy. At this stage a number of system responses can be triggered, again controlled by software switches.

As an example, automatic samples approval can occur when the method operating range is met, calibration verification is valid and the controls are in range. But still the system checks the specification and reports the resulting sample status. In other cases a specification violation can automatically generate a retest.

Results from newly approved samples are then distributed according to definitions: every time SAP R/3 or sub systems are notified, in some cases (like casting ingot) data is averaged over a certain length/time to present a realistic picture of the entire casting procedure.

Summary

Based on the special requirements of the metal industry a laboratory information management system must provide a high degree of power, flexibility and automation. WinLIMS is a modern system that delivers these. User interaction is limited to two or three forms, minimising the need for training and making the system easy to use and easy to learn for non IT-personnel.

At the same time, the system performs without significant delays on a database containing approximately one million samples and 20 million parameters, thus guaranteeing to adequately managing the high sample throughput.