Anthony Vatterott

The article, “A dynamic logistics coordination model for evacuation and support in disaster response activities”, by authors Wei Yi and Linet Ozdamar, and published in the European Journal of Operational Research, presents a robust two-phase optimization model to advance the accurate delivery of both material and human resources to and from both permanent and temporary facilities within the affected zone of natural disaster. In the model, a simulated earthquake affecting urban area of Istanbul, Turkey (which is expected to happen within 30 years of this publication) is used for example. The scenario presented outlines the several time periods throughout the initial phases of the disaster response, including the initial phases of setting up and making operational temporary facilities through which to treat light-medium to heavy healthcare needs for disaster victims, as well as the reconfiguration of optimal routing to reach optimal facility locations within the most drastically affected zones. Later in the model, time periods focus on replenishing of supplies and relocation of further human resources to appropriate facilities based on the queuing that occurs at each location.

The model developed and implemented in this paper is a mixed integer, multi-commodity network flow model that treats vehicles as integer commodity flows as opposed to binary variables. This allows the model to construct an optimal vehicle route and load instruction sheet, based on demand of various supplies along a given route. The goal is to minimize the delay in making priority commodity and human resources deliveries to affected areas. The optimization of both integer material and human resources and vehicles, and then re-calculates the optimal distribution of these resources after delivery in each time period is made. The model both makes suggestions on the location of temporary response facilities within the affected disaster area, and minimizes the transportation delay for patients with different medical treatment level requirements and at different locations. The model can adjust to regular information updates including the interruption of roadways and supply lines, the breakdown of vehicles (taking them out of the pool of availability and re-ordering priority) and re-allocation of service capacities. The model is based on the assumption that information flow after a disaster is within a planning horizon that is extremely short and initial feedback from sites is scarce, until later periods where information more accurately resembles the demands at local emergency sites.

The model is a Location Routing problem, which takes into its development aspects of both Vehicle Routing and Facility Location type problems. The model involves split-delivery, deterministic demand that is dynamic as opposed to static over a number of time periods, involves multiple capacitated facilities with size optimization requirements, and a heterogeneous fleet of vehicles (in this case, helicopters, trucks and ambulance) each with different capacities.

The Vehicle routing sub-problem deals with several special considerations. Supply availability is limited due to the disaster’s impact is difficult to immediately determine, and there exists a transportation delay from major warehouses. In addition, the ability to have in-stock levels of commodities required at the onset of disaster may not be possible, and supplies may need to be redirected from other locations to that facility first. In addition, vehicles may be redirected from one delivery endpoint to another node of delivery and pickup, and not simply back to a point of origin, making the need for the VRP to be run at each time period considering the location and needs as re-prioritized in each time period.

However, the benefits are great. The flexibility of the model, and the holistic allocation of resources and commodities to the location they are most needed in order of priority, reduces unnecessary injury complications and possible casualties as well as using scare resources where they will have the greatest overall impact.

The model is ideal for the type of unique scenarios that disasters create; however, because demand information may not be immediate and the model uses a deterministic approach to re-evaluating demand over time periods, it does not make decisions as to the pre-positioning of supplies at locations before the disaster event in preparation of the possibility of occurrence. Therefore, more efficiency may be gained by expanding the model to include the probabilistic considerations of likelihood of various scenarios, the range of labor and material demands and the location of warehouses before the onset of disaster. This would benefit the model by positioning the resources for response more within the likely best-required location at the start of the deterministic model. Because it is impossible to anticipate when or where the disaster event will occur, it makes sense to minimize the cost of the best possible pre-positioning of goods for various probabilistic scenarios. However, a tradeoff occurs in which, based on probability of a disaster event occurring, products of a high dollar value may also have a high priority value in time of actual disaster. Therefore the value of products must be weighted against storage requirements (inventory) costs as well as product purchase costs. Mete and Zabinsky provide a mathematical formulation for determining the value of goods in inventory at warehouse locations prior to a natural disaster: (Part 1) Minimize the Sum of (Cost of Operations for a Warehouse X Number of Warehouses Operating in a Pre-positioning Capacity) +A Probabilistic Value (Part 2) Based on the Demand Probability of Various Scenarios, subject to supply storage space constraints of each available warehouse, and maximum availability levels of the range of medical supplies.

The probabilistic value of demand is based on the occurrence of numerous iterations of simulated scenarios whereby the LRP uses various levels of supply and warehouse location, as well as the level of unsatisfied demand under those scenarios. What this creates is a feedback loop of optimization of warehouse location and inventory levels based on probabilistic demand, and then deterministically responds to the crises, effectively optimizing both the emergency response preparedness and the disaster response.

Other possible directions with which the original model can be developed include scaling the problem for minimum and maximum affected areas so that local, regional and larger scale disasters can be considered simultaneously with a weighted probability for each scenario to occur. Another possible consideration is the determination of vehicle capacity on various roadways affected by disaster, so as to determine the better option for a delivery route over two open roadways with different capacity flow levels (1 versus 2 lane; popularity among local residents for use during peak, off-peak and emergency hours); and work schedule assignment constraints such as maximum hours allowable for healthcare and response individuals to work before rest is required (considering the effect on service levels and service times, the queuing scenario and next best options for staffing).

The two-part LRP is a great development in the study of emergency response because it considers the time constraints and the immediate necessity of the initial phases of disaster recovery. When coupling this phase with the upstream and down stream linkages in the supply chain, as well as considering the pre-planning stage and the concurrent modifications and limitations of the local, regional or larger scale infrastructure, as well as both deterministic and probabilistic modeling within simulation and optimization approaches, the best case scenario can alleviate suffering, save lives and best allocate scarce and costly material and human capital.

References:

Mete and Zabinsky. Preparing for Disasters: Medical Supply Location and Distribution, Interfaces (2009)

Yazici, M.A. Ozbay, K. Impact of Probabilistic Road Capacity Constraints on the Spatial Distribution of Hurricane Evacuation Shelter Capacities. Transportation Research Record 2022 (2007) 55-62

Yi, W. and Ozdamar, L. European Journal of Operational Research 179 (2007) 1177–1193

Lifehacker today has an article that goes in-depth on the subject of setting up your computer using BitTorrent and other applications as a Personal Video Recorder (PVR) and interfacing this with your television.

A while back I explained how to set up your home network including your computer, television and other devices, to create an expansive home media center. Really, in this day and age of interface compatibility, there is no reason NOT to have a personal media center that allows you to stream your computer and hard drive-based media to your television.  Netflix has made this sort of interface the next billion dollar business.  And you too can specialize your own cobbled-together network at home, in order to access movies, television from hulu and other websites, and whole programming lineups for most cable television networks.

The essential components of this set up are:

-home high speed internet with bandwidth to allow streaming of large files in excess of 1mb/sec upload (the best quality of data transfer is achieved with 4-8mb/sec upload speeds or greater.

-personal computer on this home network, with processing speed and memory capacity that is standard for the last 2 or 3 years (older computers can also be used with little problem.  I actually have a generic desktop workstation from circa 2000 running software for VOIP and as a media server).

-a Wii, an Xbox, a PS3, a Boxee device (or make your own!!!), a google tv device, or a DVI to HDMI cable to interface the computer with your television.

-a software method to convey media to your television (Medialink, Vuze, TVShows, Playon).  These can range from simple media server plug-ins like Medialink that allows you to share your computer media files with other computers or devices that are Universal Plug N Play (UPnP).  Or they can be full applications with additional scripted plug-ins like Playon TV, which allows streaming of television shows from the actual origin website, including live sports events like ESPN3 directly onto your television.

Of course, Vizio–a Television brand that I highly endorse for its claim as the only television manufacturer that is American–has added the Vizio Internet Apps (VIA) interface to its higher end viewing platforms.  However, it doesn’t really give you complete control over setting up RSS-style feeds to automatically download  your favorite shows–the sort of system Hulu does so well.  By establishing your own media server you can gain access to these feeds directly from Hulu, but without having to log into the Hulu website in a browser.  You can simply use your media server to access your favorite shows.

For those of you who are Internet users, but are not quite sold on cable and the cost of all those un-used channels, the possibility of tapping into JUST the shows you want to watch from LMN, TBS, TMC, or ESPN is very tempting. I can tell you my wife and I have watched full seasons of 48 Hours Mystery, Glee and Sons of Anarchy using the home personal media server setup.

If you’re like me and you have an extra computer laying around, do some research and set up your own media server!  It can be as big or a little as you want.  The cool thing about it is the way you can shift your media consumption to the times, and shows, you want to watch, without having to sport the money for a Tivo, and without having to get cable or OnDemand, and without the extra bill every month!

I personally believe the trend is going to continue to drive the level of sophistication of setting up these media devices down to the consumer level, and soon everyone will be subscribing to their shows as opposed to just watching them during primetime.  It’s almost there to the point it is as easy to time- and place-shift your media (meaning you can access your favorite shows or music anywhere, at any time) as downloading an app for your iPhone or Android device.  Until then, a little research will help bridge the product gap.

Reposted from CNN

Shahdadkot, Pakistan (CNN) — Shahdadkot’s half-million people frantically tried to flee their homes Saturday as a wall of water threatened to burst mud berms and drown the entire city in Pakistan’s Sindh province.

Three weeks into the worst natural disaster in Pakistan’s history, people were still desperate to escape as a second wave of monsoon floodwaters surged southward. More than 1,500 people have died and 20 million lives have been disrupted.

Already, huge parts of Shahdadkot look like a lake, with the roofs of some houses barely above water. Authorities advised the entire population to evacuate.

Residents climbed onto heaps of belongings piled high in the beds of rickety trucks, packed buses, auto-rickshaws and carts to get out of town before the water came. Many did not know where they were going — just that they had to reach drier ground.

But there weren’t enough vehicles for a mass evacuation.

Sunat Magsi and her 100-strong extended family lost their nine mud huts to the raging torrents. They sought shelter in an abandoned house, but even there the water was creeping higher. They only had one donkey and one cart left.

“We have so many children here,” Magsi said, weeping. “We don’t know how we’re going to get out. We need help.”

Pakistan is dotted with villages, towns and cities submerged like Shahdadkot. Floodwaters are expected to recede in the next few days as the last surges in the Indus River flow into the Arabian Sea.

But the suffering is sure to continue. Health officials fear that the human toll will get a lot worse as people are forced to wade through unsanitary water while clean drinking water is scarce.

More than 200 health facilities have been damaged or destroyed, according to the World Health Organization, greatly reducing the available health care for millions of survivors in filthy conditions. At least 4 million people are homeless.

Dr. Guido Sabatinelli of the World Health Organization said reports of diarrhea have increased 30 percent.

Map: Flood-affected areas of Pakistan

Video: Stubborn flood victims refuse to leave

“The depth of suffering is incalculable as risks escalate of diarrhea, acute respiratory infection, malaria and other communicable diseases,” Sabatinelli said. “It is crucial that all humanitarian health providers, local and national, coordinate their relief efforts closely to save lives, reduce suffering and deliver the most effective response.”

United Nations officials have appealed for $460 million over the next three months to help the roughly 20 million people in need of shelter, food and emergency care.

The U.N. has received about $263 million and an additional $54 million in pledges — or about 70 percent of its goal, said Maurizio Giuliano, spokesman for the Office for the Coordination of Humanitarian Affairs.

WHO said waterborne, airborne and contagious diseases, including acute watery diarrhea, measles, malaria and acute respiratory infections, are threats due to overcrowding, lack of hygiene and breakage in waterlines.

Meanwhile, International Monetary Fund officials said they will meet with members of the Pakistani government in Washington next week to discuss the economic impact of the massive floods that have ravaged the country.

“The floods which have hit Pakistan in recent weeks and brought suffering to millions of people will also pose a massive economic challenge to the people and government of Pakistan,” said Masood Ahmed, director of the IMF’s Middle East and Central Asia Department.

Ahmed says the meeting will be an opportunity to evaluate the economic impact of the floods, assess what Pakistan’s government is doing to respond to it and “discuss ways in which the IMF can assist Pakistan at this difficult juncture.”