The sampling frequency and procedure should be well planned. An inadequately planned and prepared sampling programme may ruin the whole tracer project. Important aspects to consider are the following:

• Is each individual production well accessible for sampling? In most land based reservoirs this is the case. In offshore wells where the wellhead is ‘dry’, i. e. placed on a platform above sea level, this may also be the case. However, for subsea completion where the well heads are placed on the sea bed, each individual well may not normally and easily be accessible for frequent sampling. In this situation, the well flows from several wells come together and are co-mingled in one transportation pipeline from the bottom hub to the receiver installation either on an offshore platform or onshore. Individual well sampling may not be possible in this case. However, in order to enable production testing of individual wells, the bottom hub is normally equipped with valves on the production line from each individual well. Thus, by closing these valves according to a certain procedure, the production from one well may be increased at the expense of the others. In this way, the production of a certain tracer may be associated with a specific well (or a limited selection of wells) each time a production logging operation is carried out.

• Cross-contamination: When each individual well is accessible for sampling and the sampling is performed directly in the flow line, there is a very limited possibility for cross-contamination of the collected samples, i. e. that waters from various wells mix in the same sample. However, human error and erroneous labelling of the collected sample may happen and lead to confusion in the results. Most often, the collection of water samples, even from individual wells on a platform, is performed on a test separator which is common for several wells. In this case, cross-contamination can more easily happen. The remedy is to ensure that all water from testing of the previous well is swept out before sampling the next well.

• Sampling procedure: Regarding concrete sampling procedures some questions must be considered, Will it be discontinuous sampling involving personnel for each individual operation? In which case, who is going to do the job? Are there special requirements for sampling containers or stabilization additives to the sampled fluids (e. g. for preventing microbial degradation of the tracer during transportation and sampling)? Is it possible to adapt some form of automatic batch sampling so that personnel are not so involved at each operation? These are all questions that must be clarified upfront.

• Sampling frequency: How often should samples be collected? This question should be answered on the basis of a best estimate of expected tracer breakthrough. As a general rule, the sampling frequency should be relatively high in the beginning, starting shortly (a few days) after injection. All samples are stored safely. As an example, each fifth sample is sent for tracer analysis. After the first tracer detection, the previous four samples are analysed with priority reversed in order to determine more precisely the tracer breakthrough time. After tracer breakthrough, the high sampling frequency should be maintained and each sample analysed until the tracer production peak lies on a decaying slope. On the tail of the curve, the sampling can be less frequent.

The sampling frequency (samples/d) depends on the geometrical size of the tested field section and on the flow rates involved. In an average sized reservoir with well distances in the range of 500-1000 m, a sampling frequency in the beginning can be one sample each 2-4 d, followed by the same frequency after tracer breakthrough. After having passed the peak maximum, the frequency can be reduced to one sample per week or per two weeks and later even to one sample per month. It is, however, recommended that sampling and analysis be continued as long as possible, since much information from a tracer test lies in the tail of the curve. The LD value for the particular tracer sets a practical limit to the sampling period.