I don’t think life-cycle assessments are being done well. Here’s my suggestion for improving them.
A life-cycle assessment (LCA) is a kind of analysis that evaluates the environmental impact of a product or process. There’s one particular detail about how LCAs are carried out that I think is seriously flawed. I’m going to focus on that detail here. If you want more information about LCA in general, there’s all kinds of very accessible information on the Web.
LCA proceeds by breaking down every aspect of a product’s manufacture, use, and disposal, and then assigning to each aspect a certain impact. You then add up all the impacts and that’s the total impact of the product. The impact is a non-dimensional number – it has no units. And since there’s many different ways to calculate the impact values, we can’t say the impact of one thing or another is really fixed or absolute. So as an absolute measure of environmental impact, LCA is relatively useless.
On the other hand, LCA is very useful when comparing alternatives. Should you use paper or plastic? Bottled water or a steel, refillable canteen? Are we doing better now than we were before? That sort of thing. Being sustainable is about making decisions, and decisions should be based on rational comparisons of alternatives.
The problem with LCA assessments as they seem to be carried out these days, is that the impact of a thing is based on a “cascade” of all the other things that led to it. This makes the problem of calculating a good LCA value intractable and, I think, quite useless. To see what I mean, consider a simple example: a plastic bottle of water.
To get the LCA impact of a bottle of water, you need to sum the individual impacts of each of the bottle, the label, the cap, and of course the water itself. Let’s look briefly at the most obvious thing: the water itself. You’d think that the water wouldn’t have that much of an environmental impact – after all, it’s just water – but lets work through a bit of the cascade, and you’ll start to see what I mean.
First of all, did you bring the bottle of water home from the store? Did you drive to the store to get it? If so, some tiny fraction of the gas your car used to get to the store and back has to be included in the water’s impact. Ideally, it would be based exactly on your car; you’d need to know the fraction of gas used to carry the water as a proportion of the rest of the “cargo” that your car carried on that particular trip.
But even the gas has an impact. Not just in its burning in your car’s engine, but also in the environmental impact of its refinement from petroleum. That’s the cascade in action. You’d have to calculate the impact of all the refinery equipment, transportation equipment, and other resources needed to make that tiny bit of gas needed to bring the water home from the store. You’d have to include proportional amounts for all the paper in the forms used to manage the process of making that tiny bit of gas, a proportion of the HVAC needed in the offices and factories where the people worked to make sure that tiny bit of gas got to a station for you to buy, as well as a tiny proportion of the impact of the gas station that sold it to you.
If you bought the bottle of water at a store, you’d also have to include a fractional amount of impact for the store itself: the HVAC, the lighting, even the construction of the store itself. And what about the teller at the store’s checkout? That person may have driven to work, so a fractional amount of that gas he used to get to work was necessary for him to check you out when you bought the bottle of water. You’d even have to include an impact for the amount of paper and ink used to print the receipt when you paid for the bottle of water.
Then you have to do all that work again, for the process of getting the bottle of water from the supplier to the store. And for pumping the water from a source to a processing plant, and then to the supplier.
Of course, this assumes you have a reasonable basis for measuring environmental impact to begin with. Is it CO2 production? Total GHG (greenhouse gas) production? Carbon footprint? Or is it something else entirely? And just how do you accurately figure out the impact of 5 cm of paper for the receipt at the store where you bought the bottle of water?
And that’s just for the water itself. You then have to do everything all over again for the bottle itself, for the cap, and for the label.
And that’s just for a bottle of water. Imagine having to do this for, say, a building, or a city.
Kind of hard to do, eh? Indeed, this is an intractable problem. An intractable problem is one for which we can conceive a solution method, but cannot expect to ever execute it. This is the LCA cascade in action.
And yet, people do LCA analyses all the time. So how do they get around the intractability? Easy: they ignore many of the impacts, on the premise that they are too small to matter. Consider, for example, the very reputable Toyota Prius. One would imagine that a Prius is better for the environment than, say, a Hummer. Your guess would be largely based on the fact that a Hummer sucks gas like a like a black hole sucks planets, while the Prius runs on the smell of a damp oil rag. (Okay, maybe I exaggerate, but you get my point.) After all, it stands to reason that the manufacture of the two vehicles can’t be that far apart, and that actually operating the cars will constitute the majority of environmental impact. And since you’re trying to avoid doing all the work, you’d ignore the manufacture of the car, assuming that it has a very small influence on the big picture.
Unfortunately, you might be wrong. It turns out that a Hummer, by some measures, vastly outperforms a Prius. (The original report by CNW is apparently available here.) The biggest problem is the nickel in the Prius’s batteries. Some argue that the manufacture and eventual disposal of the Prius’s batteries is so environmentally damaging, that it’s actual life-cycle impact is worse than that of a Hummer. This, too, is the cascade in action.
Of course, there are also dissenting opinions too. And while the Pacific Institute has a mandate to argue for environmental protection and other generally leftie causes, one must accept that they speak with some authority on these matters. They make a pretty good case that the Prius is really as good as they say it is.
So who’s right? It’s really hard to know what’s going on, even if you’re a science geek like me. What did the Pacific Institute forget or ignore in their calculations? Probably a lot. Definitely not the same stuff that the CNW report forgot or ignored. There’s been a lot of discussion about this report (for example, in the physics forums). And since the calculations were done in two different ways, you’re really comparing apples and oranges, not Priuses and Hummers.
The point is this: while it’s easy to do an LCA analysis, there’s many ways to do it, and each will provide a different answer. So you can never really be sure which analysis is right. This means that people with vested interests will use whatever results further their cause, and not necessarily the one with the most solid scientific backing. This leads to greater confusion in the public realm, which leads to indecision by politicians. And indecisive is the worst thing we could be right now.
All this confusion arises from the inherent complexity of calculating the cascade effects in LCA.
There’s another problem with the cascade: it’s difficult to see relationships in the data because all the fractional impacts from the cascade come from so many different sources and are divided up so finely.
Try to picture it this way: imagine you have a (very large) sheet of (hopefully recycled) paper. You draw a point on the page to represent each thing that has an environmental impact. The bottle of water, your car, the gas in your car, the store where you bought the bottle, the slip of receipt paper, the clerk’s uniform, the store’s HVAC system, etc.
A possible LCA diagram for a very simple product.
Now, starting from the product in question, draw lines connecting one impact source to another, such that the thickness of the lines represents the amount of impact from that one source. For example, you draw a line connecting the bottle of water to the gas in your car. This would be a fairly thick line because it’s a rather large contribution to the impact of the bottle of water. Indeed, it’s probably more than the impact of the gas needed to truck the water from the factory to the store because (a) trucks usually run on diesel, which has less impact than gasoline, and (b) the truck would have carried a lot of water bottles and so distributed its impact more widely.
The further you get from the bottle of water, the thinner the lines will tend to be. For example, the impact of the HVAC in the factory that made the ink that was used to print the receipt at the store when you bought the bottle of water – this will be a pretty thin line. But there are exceptions: if you drove a Prius to the store, you might have quite a thick line to the manufacture of your car’s batteries.
What you’d end up with is a strange, lop-sided, starburst-like image of lines radiating out from a central point. Some lines will be thick, others will be thin.
A simple diagram of 5 products.
Now do this again for some other item drawn as a dot on your page, for instance the paper for the receipt from the store, or for the store itself. Each time, you’ll get another starburst, somewhere else on the page, maybe sharing some points with another product but maybe not. And you’ll have lots of overlapping lines.
Now imagine doing this for everything you own. Or for a building and it’s contents. Or for a whole city. Your diagram will be an unholy mess.
What useful information could you possible derive from such a diagram, or from all the reams of numbers that you’d have to calculate to construct the diagram?
And more importantly, how would you act on recognizing that 0.0000001% of the impact of the bottle of water came from a coal-fired power generating station? Would it also matter to know that 0.00000001% of the impact came from a hydroelectric power generating facility? How can you change your daily behaviour to be more sustainable, when the data you have is basically useless to you? How can politicians – who are a largely innumerate lot anyways – possibly come up with meaningful policy and regulation when this is the kind of data they are given?
A systems view on LCA
I think we can fix these problems with LCA by adopting a more systems-based approach. The key is to give up the notion of capturing the cascade, because it’s the cascade that is confusing matters. I think there are other benefits to using systems, but I’ll get to those later.
To distinguish what I’m proposing from conventional LCA, let’s call this new approach sLCA.
A system is a collection of interacting things that serve a definable function and that is crisply distinguishable from its environment. (The environment of a system is everything else except the system.) So, for instance, a paperclip floating in the infinite void is not a system. A paperclip holding two sheets of paper together still isn’t a system. But the paperclip plus the sheets it holds together do make a system.
Systems can be closed or open. Closed systems only happen in one’s dreams; they are systems entirely isolated from their environment – nothing at all passes through the system’s boundary. They’re interesting simplifications for science and engineering, but they don’t actually exist. The only possible closed system is the universe itself, and we’re not even sure about that. Open systems exchange things with their environments. In technical systems, we consider only three kinds of things that can pass a system boundary: mass, energy, and information. These are pretty broad classes, and I’ve yet to think of anything needed in a technical system that isn’t easily reduced to be of one of these three kinds.
One other thing about systems: they don’t exist in nature. That is, a system is a construct of the human mind; it is a model we use to simplify or specify something in a way we can understand and control (more or less). So where exactly we draw the boundaries that separate one system from another is entirely up to us. It would make sense to choose boundaries that line up with some very easily established characteristics of things.
For instance, we generally mark the boundary of the human body at the skin, which crisply (to our eyes) marks where our bodies end and where the rest of the world starts. Even though it’s not a very accurate model, it works for typical cases in daily life. If you could see very small things, you’d see very close to our skin is a microscopic whirlwind of exchanges between our body and the atmosphere that would significantly undermine the idea of a crisp boundary. But if you’re only interested in typical human-scale activity, it’s perfectly acceptable to mark the skin as this kind of boundary.
What we want in sLCA is to set up a series of systems that interact by sharing mass, energy, and information. We can measure how efficient those interactions are, and thus describe the environmental impact of the system. We’d want these systems to be easily recognizable, and follow certain borders that have societal as well as technical relevance. This will result in a far more natural description of the life-cycle, that will be more meaningful, and more useful, to more people.
Let’s consider a house as an example. The obvious boundary here is at the limits of the property on which the house is located. It is an obvious physical boundary – it will share well-established boundaries with other properties. Perhaps more importantly, it marks the boundary of the control that the house’s owner can exert.
This is worth describing in more detail. In conventional LCA, I am responsible for a proportion of the environmental impact of the power generation facility that provides electricity to my home. But I have little, if any, control over where the electricity comes from. No matter what the source of the electricity, it’s all one big pile of electricity once it gets into the grid. By the time it gets to my house, I have no control whatever over which electricity I use. It would be nice if I could only use the electricity that comes from Niagara Falls, and not use any of the electricity from Nanticoke. But I can’t. I have no control. And yet, I’m responsible for the cascade impacts of that electricity.
Why should I be responsible for the environmental impact of things over which I have no control? As far as I’m concerned, there is absolutely no reason at all for it. Assigning responsibility without also assigning control is not only irrational to me, but also highly unethical.
On the other hand, by putting a boundary around the house, we’re saying that the person responsible for the system is only responsible for the impacts of the system itself and not for the impacts of things that happen outside it.
It’s now a straightforward exercise to measure the mass, energy, and information inputs and outputs through the “house” system. Energy inputs include things like natural gas and electricity; energy outputs include waste heat, light, and noise. Mass inputs include pretty much everything you buy and bring into your house; mass outputs are your garbage. Information is perhaps the least relevant, because information is generally not considered to have an environmental impact. (This is not to say that the media of transmission of that information is also impact free, but the media will be either energy or mass and so are already covered.)
“Efficiency” can be measure very easily now. Electrical consumption is already measurable very precisely at the level of a house. So is natural gas. We don’t even need to convert it to a carbon footprint. We just need to look for ways of lowering consumption and raising efficiency. We don’t need to measure carbon footprint to do that.
Waste (garbage) is also easily measured in that it would be easy to set up a system to measure how much of different kinds of crap we leave at the curb on garbage day. In Cambridge, UK, for instance, they use three different kinds of recycling boxes: compostables; papers; and plastics, metals and glasses. Each requires specific recycling processes that are hugely facilitated by human pre-sorting. There’s a “real garbage” container too. Given such a system, one can measure waste sufficiently well just by measuring (perhaps weighing) the containers. Again, we don’t need to reduce the measure of waste to some abstract value like carbon footprint, because we know we want to lower consumption and increase efficiency.
The owner of the house is responsible for the house, and he has the control to take action. He might, for instance try to reduce the amount of garbage he produces by composting. If he uses his own compost on his property, some of his garbage (output) would literally vanish in the sLCA analysis. If he sells or gives away the compost, it would not appear as waste, but as a beneficial output.
This raises an interesting question: beneficial outputs are not waste products. Indeed, something like producing compost not only lowers the amount of garbage, but provides a useful input for other systems. So how do we distinguish them from the wasteful outputs in sLCA? Easy; just subtract the mass of compost output from the amount of garbage produced. The same would apply (and in fact does apply in certain enlightened parts of the world) to electricity. If you produce excess electricity with, say, solar panels, you get a double credit: you not only consume less electricity from the grid, but get a bonus for producing the stuff. (Of course, you’d have to count the impact of solar panels themselves somewhere, but it wouldn’t be at the house. More on that later.)
Anything that produces heat in a house can be thought of as part of a “heating system” that contributes to keeping the house warm in winter. Refrigerators produce a lot of heat. That’s heat that your furnace doesn’t have to produce, so you’ll use less natural gas (or whatever). Incandescent lights produce heat too. Some experts suggest there are actual benefits to using incandescent lights because of this ancillary heating effect. Of course, this only works in cold climates and not at all in warm ones.
This brings up an interesting, and sometimes confusing, point about systems. The same component can be part of many different systems. Lights can be part of the heating system and the lighting system. Many experts (including, unfortunately, the Government of Canada) don’t understand this point. Systems are about providing function, so we have to be careful when we itemize the list of things that contribute to that function. Still, it’s much simpler than trying to work out the cascade of all the lights or the sources of heat in a house.
This is something that requires further investigation. How much heat do incandescent lights produce for the electricity they use? How does that compare to the natural gas usage of a furnace? Perhaps we ought to install heat lamps in our homes instead of using furnaces. If one can save overall energy consumption with incandescent lights, how many lights do you need to get a reasonable savings?
Still, if we’re interested in conserving, then cutting down on consumption is a great way to do it. Dimmer switches are great, as are automatic sensors that turn lights on when you enter a room (and turn them off for you when you leave it). Using natural light is also very good. (How many of you have bathrooms that don’t have windows?)
Water usage is also important – though many of us don’t think about it much. Low-flow toilets help a lot. There’s a rule that estimates each person flushes a toilet 5-6 times a day. That means that each person uses about 24,000 litres of water a year, just to flush toilets (assuming “standard” 13L toilets). There’s an old chestnut from the groovy Hippie days: if it’s brown, flush it down; if it’s yellow, let it mellow. This guideline alone can save 25%-50% of the water flushed away – at the possible expense of a certain aroma. Changing to better toilets (that use 6L or less per flush) will get you the same savings.
But that’s only one side of the equation. There’s another input that can be used here: rainwater. Collecting and using rainwater, even if just to flush toilets, is a viable alternative because it’s not water coming from the public system – which will have its own impact. You save money because rainwater is free, and you’re helping your community lower its environmental impact because the water distribution system doesn’t have to work as hard.
You end up with a kind of jigsaw puzzle. Each piece is distinct and entirely self-contained. These are the subsystems: heating, lighting, etc. The pieces fit together so that there are no gaps anywhere (i.e. no unaccounted for functions). Where two pieces meet, there are exchanges of mass, energy, and information. The outer edge of the puzzle as a whole is the boundary of the house system, and the house system as a whole exchanges mass, energy, and information with other systems. This kind of representation of systems is called an architecture.
Okay; enough about your house. How does this help us at bigger levels?
Just as your house system can have subsystems for heating, lighting, water, and so on, so too your house can be an element in a larger system. The block your house is on, is a sensible unit here. Its usage would be straight sums of all the inputs and all the outputs of all the houses in it.
There is a catch here, though. Because some parts of the block are common areas, we might expect there to be systems in the block besides just the houses, systems for which no one person or group of persons in that block are responsible. Streetlights are a fairly obvious example. The streetlights would be a system within the supersystem of the municipality, which is the agent responsible for their installation, use, and maintenance. (Notice that the streetlighting system is a functional unit – it’s there to provide a specific function – and not necessarily just a collection of lights, but should also include the people who maintain them, the offices they work out of, etc.)
One of the nice things about the system perspective is that you can start anywhere in the hierarchy. Just pick something, figure out the corresponding system, then break it down into smaller systems, and build up larger systems with it.
So given our starting point of a house, we can think of it as a system, composed of other subsystems, all under the control of the property’s owner: a heating system, a lighting system, an insulation system, a structural system, a furniture system (all the furniture in the house), and so on. House systems assemble into blocks (plus other systems like streetlights), blocks assemble into communities, communities into municipalities, and so on. The resulting structure of systems that are all interacting across their boundaries defines an architecture for the supersystem.
At some point, one of the supersystems will get big enough that it will contain a power generation plant, like Pickering or Nanticoke. Nanticoke is the largest single source of GHG anywhere in Canada. In conventional LCA, its environmental impact would be distributed across all the energy consumers that use the power, effectively hiding it as a point source of GHG. In sLCA, its impact would stand out like a sore thumb, which is as it should be: if we’re trying to be more sustainable, then surely facilities like Nanticoke must be identified and dealt with.
Pickering is a different story. It’s a nuclear power station, and nuclear power typically has a carbon footprint of 1% or less of the equivalent coal-burning plant and is often seen as a favourable energy source. But it produces radioactive byproducts that have risks attached to them that GHG does not. There are issues of public education here, as well as technical ones, that have yet to be sorted out. Still, it is possible to assess the relative impact of the nuclear waste, because in sLCA, we do not include the long term impact of the waste once it has been removed from the plant. The waste would then become part of some other system the function of which would be to secure it safely.
The overall responsibility for both the nuclear plant and the waste management system would belong to the agents who have control over the supersystem that contains them both: (probably) an appropriate government agency. Which, again, is as it should be.
There’s another interesting side-effect of sLCA: it’s easier to spot opportunities for so-called industrial ecologies (IEs). In an industrial ecology, a group of entities that consume materials and produce waste develop mutualistic symbiotic relationships. In other words, the waste products of one entity are useful inputs to another entity. If you do this well enough, you can very substantially reduce the total environmental impact of the ecology because each element is essentially “feeding” off the others. There have been several successful IEs (such as in Kalundborg, Denmark). To identify opportunities for these IEs, you need to be able to identify the inputs and outputs of a group of entities. You then look for closed-loop chains consisting of organizations that need as input the wastes produced by other organizations. If you were to include the cascade information too, as in conventional LCA, you would ruin everything, because an IE is based only on the direct inputs and outputs.
So if we had an extensive and complete sLCA system architecture of, say, a whole city, it could be very easy to look for the kind of input and output compatibilities needed to form an IE. Ideally, the elemental organizations would be very close to each other geographically, but that’s mostly to limit the environmental impact of transporting stuff between them. One could start by forming any IE whatever, even if there are some transportation impacts. If this works, one can put in place guidelines that would help the formation of new IEs based on businesses and organizations moving as the get the chance or need to. For instance, if one organization is growing and needs to find a larger facility, it’s management might consider looking for locations nearer to other organizations with which it could form IE relationships.
I really can’t see how this could be done with conventional LCA, because of the cumulative nature of the cascade.
So there you have it (even though I’ve only scratched the surface of sLCA here). A systems-based LCA method should be simpler to carry out, provide far clearer indications of where the real problems are, present information that is directly meaningful to those who have the control to improve things, and leads to some extra opportunities – like facilitating the formation industrial ecologies.
I intend to make this sLCA thing a part of some of my upcoming research grants. If anyone wants updates, stay tuned to this blog.