Experiences with a GC-Computer System in an Industrial Analytical Laboratory

Erfahrungen mit einem GC-Rechner-System in einem industriell-analytischen Labor

Essai d'un ensemble chromatographes en phase gazeuse-calculateur dans un laboratoire d'analyse industrielle

J. Baudisch

Erd61chemie GmbH, D-5000 K61n 71, P. O. Box 75 2002, Germany (West)

Summary: The Siemens 300 GC-System, a dedicated, large and rather comfortable GC-System, monitoring 60 gas chromatographs is briefly descried. Prevented and unex- pected difficulties are reported with respect to hardware as well as software and operation. Although the results are satisfying, some desirable supplements are mentioned.

Zusammenfassung: Das Siemens 300 GC-System, ein grot~es, komfortables "Nut GC-System" ftir 60 GC, wird kurz beschrieben. Vermiedene und unerwartete Schwierig- keiten werden berichtet, sowohl im Hinblick auf Hardware, als auch auf Software und Benutzung. Obgleich die Ergeb- nisse zufriedenstellend sind, werden einige wiinschenswerte Erg~inzungen erw~ant.

R6sum6: On d6crit bri6vement le syst~me Siemens 300, important, largement calcul6, exclusivement utilisable pour la chromatographic en phase gazeuse et capable de traiter les donn6es fournies par 60 chromatographes. On pr6sente les difficult6s qui ont pu ~tre ~vit~es et les diffi- cult,s impr6vues, rencontrdes tant darts le domaine de rappareillage (((hardware))) que dans celui du programme (((software))) et de l'utilisation de l'ensemble. Bien que les r~sultats soient satisfaisants on indique quelques modi- fications souhaitables de l'appareillage ((~hardware~)) et de la programmation (((software,s).

e) Separated signal- and digital lines

f) Startboxes for each gas chromatograph. The Startbox contains not only a Startswitch, but Indicating-lights and an 1:16 amplifier

g) 6 input consoles with indicating-lights

11) 5 teletypewriters, three of them input + output with slow punched tape reader and puncher

i) A fast punched tape reader 200 characters/see

2. Software

The software includes three main parts:

a) On-line-data processing. This part traces the signal to determine beginning, maximum and end of a peak ac- cording to preset criteria. Simultaneously, the peak area is integrated. The peak area and the unique points of the peak are stored.

b) Real-tinae batchwise calculation. After completion of the chromatogram, this program identifies peaks, traces baseline and corrects peak-areas including allocation of peaks on tailing, multiplies these areas with the corre- sponding calibration factors and calculates the per- centages.

c) Input-output of data. This part channels the input data and arranges the results for printout of the report.

A Brief Description of our Siemens GC-System for moni- toring 60 gas chromatographs (Fig. 1).

1. Hardware

The Hardware consists essentially of:

a) A Process-Computer Siemens 305 with 16K core-memory b) A 64K drum-memory

c) A special analog-to digital-converter with 14 bit resolu- tion

d) A solid-state multiplexer (10 kHz). A 1:4 central high quality amplifier

3. Operation

At the console the operator puts in the GC-No, (if ready a green light goes on), the appropriate Job-No, the Sample- No and, if necessary, data for the internal standard. Hav- ing completed this, a flickering red light demands sample- injection and start, provided that the input was reasonable and passed the computer controls. After starting, the operator can forget about the analysis until the report is printed. In the meantime the system monitors the detector signal for each peak. Unique points of the peaks are deter- mined and stored together with the peak areas. After the chromatogram is completed, monitoring stops and the results are calculated from the compiled stored data and the report is typed.

Chromatographia 5, 1972 Official Papers 79

In terface

J G( Multi-

plexer

Digital-

Clock lOHz

t I Analog i I

I ' ~

G C Digital I t Converter i

Startbox

Fig. 1

�9 Design of GC-Computer System

�9 Schema eines GC-Computer-Systems

�9 Sch6matisation de l'ensemble chromatographe en phase gazeuse-calculateur

r

F- -fl I I

Program Pattern

Data Acquisition

Peak Detection * Data Reduction

+ Data Accumulation § Storage

t Area Correction

t Peak Identif ication

I Analysis Calculation

I Data Arrangement for Report

LL I - I

-• Job- No- Input

=1 Demand. Console Jnp u t

Extension

Memory

B Difficulties

1. Prevented Difficulties

la) Hardware

Having learned from others, an installation of separate GC-mains and grounding together with a thorough inspec- tion of the GC's saved us additional troubles.

In detail, we installed a separate transformer to supply the computer and every gas chromatograph by single, indivi- dual lines. Likewise we provided a reliable grounding with individual ground-lines to the computer and every gas chromatograph, taking care to avoid groundloops. The shielding of the analog lines is twofold: the inner shields are connected with the computer grounding only, whereas the outer shields are grounded either at the gas chromato- graph or the computer end. Which one had the best effect was determined individually by experiment. The gas chromatographs have been individually checked for the noise level at signal output, having the philosophy that any noise is preferably suppressed at its origin. It was necessary to make minor alterations within the circuitry of the instruments mainly providing direct outputs of the detector signal, circumventing the attenuator. If necessary, capacities and inductivities were installed to suppress noise from relays or thyratrons of the temperature control

system. In some instants it was regarded necessary to alter the circuitry of the gas chromatograph in order to bring all different parts of the instrument to the same ground level.

lb) Software

Since the software has been developed according to our suggestions, we tried to set these not only to get the highest possible precision, but also to make the system foolproof and easy to operate without limiting its flexibility. This implied essentially two items: an elaborate peakfinding and integrating system including a variety of different, changeable parameters, together with the appropriate hardware; and a console-input with a minimum of nec- essary input data.

Consequently, our Job-No. not only defines the retention time list and calibration factor list to be used, but also a set of 20 parameters for peak finding, integration and calculation. Additionally the heading of the report is also selected by it (see Table II).

A variety of logical controls ensure the greatest possible fool-proofness.

ic) Operational

In addition to a thorough training of technicians and opera. tars, repeated explanation of the meaning and aim of the

80 Chromatographia 5, 1972 Official Papers

system together with motivation helped to prevent an anti-computer mood.

Automatic statistical calculations on results; and

Computer control of the gas chromatographs involving mainly column-switching and temperature programming.

2. Unexpected Difficulties

2a) Hardware

Besides two failures o f electronic parts of the computer within two years, we encountered no serious hardware difficulties. We credit that to the careful design of the lines as well as to the cleanup o f our gas chromatographs in combination with the separate power supply and ground- ing

2b) Software

In contrast to only few hardware difficulties, it took con- siderable time to eliminate all the major and minor errors of the program itself. Particularly the tracing of errors in the program parts was tiresome.

Moreover, it took some time learning to optimize the dif- ferent parameters, such as thresholds for first and second derivatives, time-windows, baseline drift, scan rate etc. Particularly it was not always easy to comprehend the con- sequences of the different parameter-settings.

2c) Operational

Besides the above mentioned necessity to learn the appro- priate settings o f the parameters, which o f course is more or less operational, we encountered no serious difficulties in everyday use due to the easy manipulation and the thorough training of the operators. In starting the whole system there have been some, mostly psychological, in- hibitions on behalf of the operators, but, by pointing out that the computer serves essentially as a multiple integra- tor without any drives of its own, this uneasiness was gradually converted to the pride of being appointed to work with such a modern device.

3. Desireable I tems

3a) Hardware

Although the hardware is doubtless of high standard - proved by an extremely good reproducibility - some wishes remain. Presently, our main concern is the rather limited extension memory of 64K, since the comfortable program together with various retention tables, calibration factor lists, parameter lists, etc. require most of it. Of course, there is no difficulty to replace it by a larger ex- tension memory (512K).

3b) Software

Despite o f several features, that make operation simple and largely foolproof, some functions are not yet included, due mainly to the restricted extension memory. These are:

Calculation o f a combined analysis report from several runs on different columns;

Nonlinear peak area reallocation o f fused peaks or peaks on tail;

Control of results being within preset limits (e. g. specifica- tions);

3c) Operational

Presently there seem to be no wishes left, the handling being as simple as possible.

Table I: Repeatability

1. Packed column, thermal conductivity detector, area normaliza- tion Sample size approx. 2 ,ul

concentration relative standard of components deviation 10-100 % 0,1-0,2 %

1- 10 % 0,6-0,7 % 0,1- 1% 1,3-3,2% less than 0,1% 9 %

2. Capillary column, flame ionization detector, area normalization Sample size approx. 0,01/~1 (by splitting)

concentration relative standard of components deviation 10-100 % 0,2-0,4 %

1- 10 % 0,4-0,9 % 0,1- 1% 0,9-3,9%

3. Packed column, flame ionization detector, internal standard Sample size approx. 2 ~1

average concentration relative standard of components deviation

32,6 ppm 2 % 30,8 ppm 1,5 % 10,4 ppm 5 %

851 ppm 0,5 %

Table I1: Parameters for a Job

1. Job. No. 2. Heading of report (e.g. "benzene pure" or "ethanol crude") 3. Mode of calculation (e. g. normalization, internal standard etc.) 4. Number of known components 5. Threshold first derivative 6. Threshold second derivative 7. Minimum peak area 8. Minimum peak height 9. Minimum peak width

10. Time of scan rate change during run 11. Suppression of solvent peak 12. Suppression of high boilers 13. Duration of chromatogram 14. Expected maximum number of peaks (including unknowns)

(There are no disadvantages of an overflow) 15. Maximum baseline drift 16. Initial scan rate 17. Limits of baseline before starting 18. Area allocation method 19. Factor for unknown peaks 20. Number of inputs at console, i. e. for internal standard etc.

In addition to these parameters, for every component including the reference peak name, calibration factor, retention time and retention time-window have to he listed.

Chromatographia 5, 1972 Official Papers 81

C Advantages a) Hardware

The hardware, especially the interface, combines high re- liability with unusual good reproducibility. Therefore we can make full use of - and are only limited by - the re- produc~ility of the gas chromatographs which is - under favorable conditions - better than normally expected. Moreover, we feel, that amplifying weak analog signals directly at the instrument prevents distortion of the signals on the lines. (See Table I)

b) Software

There should be no doubt, that integration by a computer is more precise than by an integrator, since computers may use more criteria for peak detection than only the slope (first derivative). However, in our system a lot of criteria

have been included not only to determine the exact sin- gular points of a peak, but also to distinguish actual peaks from spikes and other interferences. (See Table II)

c) Use

Due to the elaborate software the use became very simple. It is foolproof due to a variety of cross-checking and logical controls, so that it can be used by unskilled operators

Literature

11. -D. Papendick; Erd61 und Kohle 24 (1971) 10, S. 637/644 W. GoM andH. -D. Metzger; Siemens-Z. 43 (1969) Nr. 7 H..17. Metzger; Techn. Mitt. Haus tier Technik 63 (1970) S. 259/261

Received: Nov. 5, 1971

82 Chromatographia 5, 1972 Official Papers