Case Studies

Case study 1:

A large petrochemicals company in western India has a 55 km long cross-country pipeline to carry multiple petroleum products from two oil refineries to its petroleum complex. After treatment of the fraction, another 55 km long parallel pipeline carries the product back to the same refineries. The pipeline is used to transport different petroleum fractions through the same pipeline as batches. The terrain is mountainous.

The pipeline was designed to carry the petroleum product at a maximum flow rate of 200 m3/hr and was operational round the clock. The company had a Leak Detection System (LDS) installed in their control room. It was based on a simple volume balancing method of leak detection. The existing LDS made unrealistic demand that the pipeline must be operating at steady state for several hours on either side of the leak event. The company desired to have our LDS which is transient simulation based and has tolerance for operating transience.

PAnORaMA LDS was adapted to suit the SCADA system which the company had for the operation of the forward and the return pipelines. Our system did not have a batch tracking feature necessary for such multi-batch systems. This was developed and incorporated in the software. The LDS was installed in the control room and shared pressure-flow data from SCADA polled at 1 minute interval. Our LDS was first tested off-line using the historical data of actual physical leak created by the company in their pipeline to test the other LDS they had procured. Our LDS could predict the leak location within ± 0.5 km. It could also predict the leak flow rate within ± 5% of the actual leak.

Some of the salient features of this project are summarized below:

1. 55 km pipeline with uniform 200 mm diameter throughout.
2. Flow rates were monitored at both the entry and exit ends of the dedicated pipeline.
3. Pressures were monitored at 4 intermediate locations in addition to the inlet and exit ends. There were thus 6 pressures available along the pipeline.
4. The pressure and flow data was polled at 1 minute interval.
5. PAnORaMA LDS had to be modified to incorporate multi-batch transport. Batch tracking feature was included in this specific adaptation.
6. The batches had different properties (density, viscosity) and hence different hydraulic characteristics.
7. Colebrook-White correlations were used for pressure drop calculations.
8. Bernoulli's equation allowed incorporation of hydrostatic head at different nodes of the pipeline which ran through a mountainous terrain.
9. Our LDS worked on a separate computer in the control room network.
10. The leak detection capability was tested using data generated by creating a physical leak.
11. The LDS successfully identified the leak event, leak location, and leak flow rate within 5 minutes of the leakage event.
12. The LDS made no special demands such as additional pressure monitoring or upgradation of instrument accuracy etc. It worked on the operating parameters monitored routinely by the operator.
13. The LDS console was designed to mimic the same look and feel as the display of the operator console.

Case study 2:

PAnORaMA LDS was deployed on the world's largest and most complex District Cooling System (DCS) in Qatar. The network involved a Chilled Water Network (CWN) which carried chilled water for refrigeration to different domestic and commercial buildings located within 8 km radius of the chilling plant. The CWN was a complex loop network involving pipe sizes from 1600 mm to 50 mm diameter. The total length of the piping was about 15 kms. There was pressure and temperature monitoring at each of the nodes (users of the network). The pressure, flow and temperature were also monitored at the entry of chilled water from the chilling plant into the network. The total nodes or locations where pressures were monitored were about 50 at the time of installation of our LDS. As more users get added to the network as new buildings get commissioned in the developing city, the network scope also increases.

Parallel to the CWN, a Return Water Network (RWN) also runs which collects hot water from the buildings and returns it to the chilling plant. This network is almost identical to the CWN as far as its layout and structure is concerned. However, it is different in the sense that it has water entering the network at all the user nodes and exiting the network at the chilling plant node. In terms of PAnORaMA LDS implementation, all START nodes of CWN become END nodes in RWN, all END nodes in CWN become START nodes in RWN, and similarly, all SPLITTER nodes in CWN become MIXER nodes in RWN.

The LDS was provided separately for the CWN and RWN. The DCS operating company also had divided the CWN and RWN into two different operating zones catering to customers in the Eastern and Western halves of the city. The reason to do this was operational ease and the bifurcation happened at the chilling plant itself. There were thus four different networks, namely, CWN (East), CWN (West), RWN (East) and RWN (West). All were served by a single common SCADA system. However, the leak detection system had to be provided for each network and we have 4 operational Leak Detection Systems in the control room.

The installation has been tested thoroughly and runs in real time.

Some of the defining features of the system are summarized below.

1. Very complex network involving 30 km of pipeline and about 20 distinct pipe sizes in the range 100 - 1600 mm diameter.

2. The network has pressure and flow monitors at all user nodes totalling about 50.

3. Total flow rate of water handled by the network is of the order of 44200 m3/hr (195000 US GPM).

4. The district cooling system is designed for 130000 RT (Refrigeration Tonne) capacity.

5. The DCS had inherently transient operation as refrigeration needs of individual users change from season to season, hour to hour, minute to minute literally. Accordingly, flows of chilled water drawn by the building level heat exchangers changed all the time.

6. The company had done extensive search on the kind of LDS that would suit their complex and inherently transient network and realised that they need an LDS based on our approach.