Statement of Mark Westhoff, Colorado Interstate Gas

MR. WESTHOFF: Thank you. We move from 30,000 feet down to the roadway, and now we'll go a few feet below the surface of the earth, and we'll move from transporting millions of human beings from point to point to transporting trillions of cubic feet of natural gas from the production fields into the markets where it will be used.

We have many of the same objectives. In the natural gas industry, our three-fold objective is to ensure the safety of persons and property, to maintain reliability of service, and to achieve both of those objectives in a cost-effective way.

I'd like to give you an overview of the interstate natural gas pipeline business, sometimes known as the transmission business, and I'll use the terms synonymously. I'll follow that with an overview of the systems that we use to monitor those pipelines and control them, and, finally, review some of the many uses that we put this operational data to.

Transmission systems vary significantly in their design and their operations, but there are elements that are common to all. They're spread over a wide geographical area, and they are positioned between gathering systems and distribution systems.

Gathering systems are relatively small-diameter pipe operating at low pressures, collecting small volumes at individual wellheads, bringing them to a central point, processing the gas and compressing them into the high-pressure transmission system.

The gas is then transported into local distribution companies who often don't have much compression in their systems, if any at all, and also run in relatively small-diameter pipe to deliver low-pressure gas directly to the end users, business users.

These systems are also inter-connected with one another, the transmission systems are, and we bring in a significant quantity of gas from inter-connecting pipelines from as far away as Canada, and we will deliver that gas perhaps into the mid-continent for further delivery even on to the East Coast when economic conditions arise through other inter-connecting pipelines.

The basic components of the transmission system, of course, involve the pipe. Pipe usually is 20- to 36-inch in diameter with wall thicknesses between a quarter of an inch and a half an inch typically, some smaller, some a little more in depth. The operating pressures, typically between 300 psi and 1,440 psi.

We use valves to modulate flow, redirect flow, and to isolate segments for maintenance and repair. Compression is used to overcome the frictional losses of the pipe, and this compression is typically one of two types. It is either reciprocating or centrifugal. It is fueled typically by natural gas, sometimes by electricity, and a common element to many pipelines of our size anyway is our storage fields.

These are used for seasonal storage of natural gas as well as for peak shaving purposes. They consist of depleted oil fields, reefs, aquifers, caverns, domes, and they provide a significant amount of operational flexibility and reliability to our customers.

The operational data with which we are most interested obviously involve pressure, to ensure safety against an over-pressure situation, and also to ensure reliability to make sure that we've met the minimum delivery pressure obligations that we have to our customers.

Flow rates are important to ensure that plans are being followed, and that people are receiving their entitlement and not receiving more or less than they are entitled to.

Composition is important because not all natural gas is the same. It varies widely in both thermal quality as well as the contaminants that may be contained.

And equipment status is also important for determining whether or not the system is configured to meet operational objectives. This would include things such as valve status and engine status as well.

Like the pipelines, the supervisory control and data acquisition systems vary widely in both design and in their implementation, but there are similar components in every one of those systems.

We begin at the pipeline with the transducers taking the various measurements, and typically brought into an RTU which has grown increasingly powerful through the years. A lot of flow calculations are done locally on site and recorded and maintained there for an audit trail.

It's important to note that these instruments are placed in a manner which is facility-driven; that is to say, we put them at receipt points, at delivery points, at compression, that type of thing. There will be clusters of instrumentation, and then there will be long stretches of the pipeline which have no instrumentation whatsoever because there is nothing happening, no processing, no compression, no receipt or delivery.

This information is conveyed to -- by some means of communication, and this varies across the board. We use a company owned and operated, Microwave System, in combination with radios. Satellite technology is used. Cellular phones are used as well as lease lines, and pipelines from other locations typically travels through some sort of repeater, and then we bring the data to field locations to make that SCADA data available to our field personnel for their own operational uses, and also from that point, we then take that data or a subset of that data, typically not all of it is captured at the site, but that which is pertinent to the operation of the entire transmission system, we bring that data in from all the various field locations, and we make it available to our pipeline control center, which is centrally located.

This particular architecture is conducive to maintaining a degree of redundance in the event that you lose any particular location, you've only lost that location and not the entire system, and similarly if you've lost the central computing capability, you still have some local distributed processing capability in the field and can overcome some of the difficulties associated with communication of data.

With so much data being presented to our pipeline operators at such a rapid pace, thousands -- tens of thousands of data points being pumped in every two to five minutes, data presentation is a key to the effective use of that SCADA system, and in the beginning, we were pretty unimaginative and were more involved with the efficient use of screen space, and, so, we used tables of data, but they have their limitation.

We have found that a more effective approach is to put this data up on the screen in conjunction with facility schematics to provide the operational and spacial context to the data that assist particularly new operators in assessing the internal consistency of the data and better making use of that data for operational decisions.

Trends are also important. Again, with so much data being pumped in so rapidly, there is a tendency to lose the forest for the trees, and the trend re-emphasizes to the pipeline operator that he's dealing with a continuum of operation, that what happens next depends on what happened before, and breaking through that mentality of just dealing with the here and now is aided in great degree by using just simple trends.

Using software that is running in conjunction with the SCADA system, we also do real-time analyses that the software may be running as part of the SCADA package or in conjunction with it.

One of the things we use the data for are system balances. We take thousands of pressure readings across the entire pipeline system, and with that information, combined with the pipeline design, we are able to calculate what the inventory of the gas is in the pipe, and that can be rather significant.

In the 3,500-mile pipeline system, such as our own, that approaches two billion cubic feet of natural gas, of which perhaps five to 10 percent of it is useful for short-term storage, packing and drafting as we call it, and by combining that with aggregated receipt and delivery points, we're able to do a real-time system balance to ensure that the pipeline is moving in the directions we want it to move.

Gas blending is also of importance to us because, as I mentioned, natural gas varies widely in its composition, and when a processing plant goes down, and they are no longer able to meet the specifications to protect the pipeline and deliver into our customer systems, we have to take some sort of action.

In the past, it was a simple shut-off. You just simply shut them down. But if you can keep them running, you can greatly facilitate that producer, that processor's capability in remedying the situation, and you can do that by blending it. If you've got additional blend stock coming in that is of sufficient quality, you can keep that processing plant on line blending its gas with others to make sure that you meet the reliability requirements but also facilitate that operator's return to service.

We also use software to assist us in ascertaining pipeline performance, and I'll give you an example of that. Pretty straightforward. A piece of pipe. We have several measurements taken which include the upstream pressure, the downstream pressure of a segment, and perhaps we're moving a hundred million cubic feet through that.

We will run some software, typically a steady state, but it could be a transient model, would be run in conjunction with this to come up with a calculated pressure, let's say in this case, of 800 pounds. That would be assuming that the pipe was clean and was of new quality, relatively low roughness, and when you have a deviation such as this, the sensitivity of that deviation of our operations to that deviation is really dependent upon the usefulness of this line.

If the quantity of gas that can move through this pipe is 200 million, and we're moving a hundred, you know, 15-pound difference doesn't make much difference, but if the capacity of that line is a 105 million or something, and we are approaching the capacity of that particular piece of pipe, that 15 pounds can be crucial to meeting our delivery pressure obligations.

So, we can use this software to assist us in determining cleaning schedules for these pipes, and also to assist us in more severe cases to actually alert us to the possibility of a blockage formation, perhaps a hydrate formation which is beginning to freeze. That type of deviation can be trended, and you can diagnose it more readily.

Another aspect of an effective SCADA system is alarming. We have pre-set alarms that we use and have used for quite some time. These basic alarms consist of the high and critical high alarms, low, critical low alarms, and also rate-of-change alarms when data is fluctuating either too frequently or too great a degree. We issue these alarms to draw the controller's attention to these operations.

The problem that you have here, and it's been mentioned several times, is if you put too many alarms out there in front of the gas controllers, they're going to start to trivialize all alarms, whether they're important or not, and, so, you must go to extra lengths to ensure that you're only putting out the alarms that require specific action, and, so, we've taken to employing what I call more advanced alarms.

Some of these are still relatively simple. The conditional alarms, kind of an if then else construct, which basically says that you have a multi-point conditional alarm issued when, for instance, you have a particular point which is giving you a contaminant which is undesirable, and the flow from that point is relatively high. You might issue an alarm at that point to draw the operator's attention.

However, you wouldn't do that if there was no gas flowing at that particular point. So, we can avoid nuisance alarms in that way, and then, finally, the analytical alarms, which are very similar to the ones that I've just described in the pipeline performance aspects. We use those alarms to assist us in determining whether a pipeline is performing to our expectations.

In the area of leak detection and response, real-time models have been proposed for years as a way of increasing the sensitivity that we have to leak detection or for leak detection. They are not widely implemented for a variety of reasons because there are a number of factors which conspire to decrease the sensitivity of the models to detecting leaks, one of which is, I mentioned earlier, the clustering of SCADA data and long stretches without it, and the compressable nature of the fluid also conspires to make it difficult to improve on the leak detection capabilities.

So, most of our leak detection is done either by land inspection with flame ionization or aerial inspection where they would simply spot a discoloration of vegetation over the right-of-way which might indicate a small leak.

But in the case of a pipeline rupture, which is a very extreme case, typically caused by unreported third party damage, it's usually a race between detection of that incident by way of phone or SCADA system.

Here we have a looped line which has suffered a pipeline rupture. All of the valves in this case are actuated. That is not typical. We do that in areas of relatively high-population density in close proximity to a pipeline. We will automate these valves, and these are very noisy affairs.

When you have natural gas escaping at a thousand psig going up to atmospheric pressure, it's like a jet engine, and it draws a great deal of attention. Typically a farmer with a cell phone in his north 40 is telling you you're blowing dirt just about the time you've developed the trend to spot that problem.

You will then dispatch the local field personnel to the area to get a positive ID on the location, and then you will direct the isolation operations, which can be done with SCADA. In this particular case, the operators can be told which valves need to be closed, and we can close the cross-overs and the main line block valves to isolate that segment to prevent the escape of the gas and then reroute that gas into the other pipeline.

The next phase is to simply reduce the volume of gas in the system, so we have to go to our scheduling routines to determine how much gas that we can move through the crippled piece of pipe, and then we can effect repairs.

Accident prevention in the pipeline business is largely an off-line effort, and it is very extensive. It is done largely without SCADA, and we concentrate our efforts on education, getting people to use the One-Call System to identify underground utilities and the like.

We also are very aggressive in surveillance, both by air and land, as well as internal inspection. We insert devices into the pipeline, push them down the line, and they determine whether or not we have any anomalies in the pipe wall, either owing to unreported third party damage or to some type of corrosion that goes on, and, lastly, we also are very, very involved in protecting the asset through corrosion prevention, putting corrosion-preventive wraps on the pipe as well as employing cathodic protections throughout the system.

In the area of risk data, we use -- in certain areas that are prone to soil movement, it has been used in the industry. Strain gauges placed on or near the pipe to determine when soil movement is such that undue stress is being put on the pipe, and action can be taken prior to any incident developing, and there's additional research being conducted right now in the area of acoustic sensors as well as fiber optic mats which can detect discontinuities involved with unauthorized excavation, those sorts of things.

Should those become technologically feasible and economically feasible, those will undoubtedly be put into the SCADA system for action by the pipeline operators' consideration.

Finally, I'll get into the business processes to which operational data are put. The SCADA system brings in the data from the field and presents it to the gas controllers via the displays that we've talked about earlier. The pipeline operators take an active role in the business process at this point, even unknowingly, because we share the same instrumentation that is used for custody transfer and billing purposes, and, so, when they spot a measurement error, not only is it an operational concern, but it's also a billing concern way on down the line.

It's obviously not the pipeline operator's concern that bills may not go out correct, but their quick action, their spotting of these measurement problems prevent retroactive adjustments on bills and lends some efficiency to the entire business process.

We also utilize the SCADA system, instead of putting in field data, we'll put in model data into the same SCADA system that is used to present data to the gas controllers. Those models will then put in calculated results for training purposes and thereby serve a dual-fold purpose.

We can train our operators in pipeline operations as well as the SCADA systems they'll use to operate that pipeline. That data also goes to archive or historical database or relational database, whatever you want to call it, and is made available throughout the company in pretty much within an hour or so of its collection. It's available for analysis by any groups in the company.

We in the operation side of things are one of the prime users. One of the cases in point is the development of a load forecaster, correlating your weather-sensitive demand to weather parameters and coming up with statistically-based load forecasting techniques.

We also use that data to determine scheduling quantities. It's fairly straightforward to determine what the theoretical capacity of a pipeline is, but it's a little bit more difficult to determine what the sustainable capacity is because not every pipeline can operate 24 hours a day, seven days a week, at 100 percent capacity. Engines go down unexpectedly, both on your system and those that you connect with, and, so, sustainable capacity typically is somewhat less than the theoretical capacity in determining a prudent level at which to schedule your pipeline is important in terms of maintaining adequate balances with your customers.

And, finally, in design, again I've mentioned modeling several times throughout. The operational data is indispensable in the development of models which are used to assess the impact of designs on existing demands as well as the enhancements that those system designs may have for future operations.

So, from the design phase through the operations phase and on into billing and invoicing, operational data is put to considerable use at natural gas companies.

Thank you.